Isolation and Characterization of Lectins from Vicia villosa TWO DISTINCT CARBOHYDRATE BINDING ACTIVITIES ARE PRESENT IN SEED EXTRACTS*

An uncharacterized lectin from Vicia villosa seeds has been reported to bind specifically to mouse cytotoxic T lymphocytes (Kimura, A., Wigzell, H., Holmqu- ist, G., Ersson, B., and Carlsson, P., (1979) J. Exp. Med 149,473-484). We have found that K villosa seeds con- tain at least three lectins which we have purified by affinity chromatography on a column of immobilized porcine blood group substances eluted with varying concentrations of N-acetylgalactosamine and by anion exchange chromatography. The three lectins are composed of two different subunits with M, = 35,900 (sub- unit B) and 33,600 (subunit A), estimated from their mobility on sodium dodecyl sulfate-polyacrylamide gel electrophoresis. Sedimentation equilibrium analysis suggests that the purified lectins are tetramers. They have been designated B4, A,, and A2B2 to indicate their apparent subunit compositions. The purified B4 and A, lectins contain 6.7-9.8% carbohydrate by weight; in addition, both are rich in the acidic and hydroxylic amino acids and lack cysteine and methionine. The A, lectin agglutinates A erythrocytes specifically and binds to AI erythrocytes (273,000 sites/cell) with an association constant of 1.8 X lo7 M-’. Although a blood group A agglutinating activity was recognized in the original preparation

group A agglutinating activity was recognized in the original preparation of K villosa lectins, lectins with this activity were obtained in relatively small amounts from seed extracts. The predominant lectin in V. villosa seeds, B4, does not agglutinate A, B, or 0 erythrocytes.
Lectins of defined specificity have been used over the last few years in probing the structure and distribution of cell membrane carbohydrates (1). In 1978, Kimura and Wigzell (2) reported that a surface glycoprotein on mouse T lymphocytes appeared after immune activation by major histocompatibility complex alloantigens or polyclonal activation with concanavalin A. Expression of this glycoprotein was shown to correlate both in time and extent to the levels of cytotoxicity generated. This glycoprotein, which appeared to be selectively expressed on Lyl-2' cytotoxic T lymphocytes and had M, = 145,000, was designated T145. Kimura et al. (3) subsequently reported that a lectin preparation from Vicia villosa seeds bound specifically to this glycoprotein. Furthermore, they were able to fractionate allogen-and mitogen-activated blasts into highly cytotoxic and noncytotoxic cell populations using a V. alloimmunized lymphocyte populations induced in vivo and in vitro and found that T145 was expressed on some, but not all, of these populations. No correlation was observed between the presence of T145 and the lytic activity of the populations studied. MacDonald et al. (5) analyzed the interaction of V. villosa lectin with T lymphoblasts using flow cytometry and were unable to demonstrate any appreciable difference in cytolytic activity between lymphoblasts that bound large or small amounts of V. villosa 1ectin.Braciale et al. (6) reported similar conclusions. Thus, both the selective expression of T145 on cytotoxic T lymphocytes and the specific interaction of these cells with a lectin from V. villosa seeds have been questioned.
The carbohydrate specificity of the V. villosa lectin preparations used in the preceding studies has not been elucidated. Kimura et al. (3) prepared the lectin by affinity chromatography on a column of human blood group A substance coupled to Sepharose 2B eluted with N-acetylgalactosamine, the blood group A determinant (7). In addition, N-acetylgalactosamine blocked the binding of cytotoxic T lymphocytes to the V. uillosa affinity adsorbent used to fractionate mixed lymphocyte populations. Hence, the lectin has been referred to as Aspecific, although its A1 erythrocyte agglutinating activity has not been documented. It is interesting that a number of other lectins with specificity for terminal N-acetylgalactosamine residues apparently did not exhibit the high selectivity of V. villosa for T145 (3).
We undertook the isolation of the V. villosa lectin in order to characterize its carbohydrate binding specificity and to examine the oligosaccharide structure of the glycoprotein T145 to which it was reported to bind. In the present investigation, we report the presence of two distinct carbohydrate binding activities in V. villosa seed extracts. One is a potent AI erythrocyte agglutinating activity, but lectins with this activity were obtained in relatively small amounts from seed extracts. Most of the lectin in V. villosa seeds does not agglutinate A1 erythrocytes but agglutinates erythrocytes having exposed N-acetylgalactosamine residues a-linked to serine or threonine residues in their surface glycoproteins (8, 8a).

Isolation of Lectins from V. villosa Seeds by Affinity
Chromatography-Lectins were isolated from the seed extract by adsorption to the porcine blood group substances affinity column and elution with 1 and 10 mM N-acetylgalactosamine, as shown in Fig. 1. A small amount of additional protein was also eluted with 100 mM N-acetylgalactosamine (not shown). The A , erythrocyte agglutinating specific activity in the peak fraction of the 10 mM N-acetylgalactosamine eluate is 2000 hemagglutinating units/mg, 40-fold higher than that in the peak fraction of the 1 mM eluate. The electrophoretic profiles of the seed extract applied to the column and these peak fractions are shown in Fig. 2. SDS-PAGE of the N-acetylgalactosamine eluates resolves 2 protein bands with estimated M , = 35,900 and 33,600. These have been given the designa- tions B and A, respectively. The lower band is more prominent in the profile of the 10 m N-acetylgalactosamine eluate, suggesting that it represents the lectin subunit with AI erythrocyte agglutinating activity.
Discontinuous polyacrylamide gel electrophoresis in the absence of SDS was performed on the 1, 10, and 100 mM Nacetylgalactosamine eluates. The 1 m N-acetylgalactosamine eluate contained protein which migrated as 1 broad band (Fig. 3A). In addition to this protein, the 10 mM N-acetylgalactosamine eluate contained a t least two major proteins, both migrating faster than the protein in the 1 m eluate (Fig. 3A ). The 100 m N-acetylgalactosamine eluate also contained two major proteins, although another protein which migrated just below the faster of the two was also often present ( Fig. 3 A ) .
T o determine the subunit structure of the V. villosa lectins resolved by nondenaturing gel electrophoresis, protein was eluted from the adjacent unstained gel and subjected to SDS-PAGE, as described under "Experimental Procedures." The gel lanes in Fig. 3B are numbered to correspond to the bands in Fig. 3A. The broad band in the 1 m N-acetylgalactosamine eluate, labeled 1, represents a lectin composed entirely of subunit B. In the electrophoretic profile of the 10 m Nacetylgalactosamine eluate, the upper of the two major bands, labeled 2, appears to represent a lectin composed of both subunits, with A being more prominent than B. The lower of the two major bands, labeled 3, represents a lectin also composed of both subunits, with A and B being equally prominent.
In the 100 m N-acetylgalactosamine eluate, the upper major band, labeled 4, represents a lectin composed entirely of subunit A. The lower major band, labeled 5, represents a lectin composed of both subunits, with A and B being about equally prominent. The band just below the lower major band, labeled 6, appears to represent a lectin composed of both subunits as well, with B being somewhat more prominent than A. Thus, it appeared that V. villosa seeds contain a t least three lectins corresponding to band 1, band 4, and band 3 (and 5) composed of the two subunits, A and B, resolved on SDS-PAGE. Subsequent experiments, discussed below, have shown that the intact lectins are tetrameric, and therefore they have been designated B4, &, and A2B2, respectively. Table I presents a quantitative summary of the isolation of lectins by affinity chromatography. In this preparation, 200 ml of seed extract were divided into equal parts (I and II), and A B -. I the two parts were applied sequentially to the affinity column.
The affinity column was eluted with 5 m (part I) and 2.5 mM (part 11) N-acetylgalactosamine before elution with 10 m~ Nacetylgalactosamine in an attempt to improve the AI erythrocyte agglutinating specific activity of the lectins in the 10 mM N-acetylgalactosamine eluates. After the column was eluted with 5 mM N-acetylgalactosamine, only 0.4 mg of protein remained to be eluted with 10 mM N-acetylgalactosamine, although the A, erythrocyte agglutinating activity in the 10 mM eluate increased. On the other hand, elution with 2.5 m N-acetylgalactosamine displaced only 0.6 mg of protein from the column and failed to increase the specific activity of TABLE I Isolation of lectins from V. villosa seeds V. villosa seeds were treated as described under "Experimental Procedures." The seed extract was divided into equal parts (I and 11) and applied to a column of porcine blood group substances coupled to Affi-Gel 15 (-4 mg/ml). Part I was applied to the column and bound material was eluted sequentially with 1, 5, and 10 m N-acetylgalactosamine. The column was then washed with PBS and part I1 was applied. Bound material was eluted sequentially with 1, 2.5, 10, and 100 mu N-acetylgalactosamine. Column fractions were pooled and dialyzed extensively against PBS before protein concentration and AI erythrocyte agglutinating activity were determined (HU, hemagglutinating unit as defined under "Experimental Procedures").  Lectins from V. villosa Seeds the 10 mM N-acetylgalactosamine eluate substantially. Recovery of hemagglutinating units in this preparation was 51%, although recovery in other preparations has been somewhat higher (61-70%). The yield of lectins in this preparation was 36.4 mg. In other preparations, reapplication of the run through material to the porcine blood group substances affinity column or to a column of N-acetylgalactosamine coupled to epoxy-activated cross-linked 4% beaded agarose (Selectin 5) revealed that additional protein bound and was eluted with 1 mM N-acetylgalactosamine (data not shown). With such repeated applications, the total yield of lectins was 75-80 mg/ 100 g of dry seeds, a value very similar to that reported by others (3,5,12). The N-acetylgalactosamine eluates from this preparation have been used to further purify the B4, A,, and A2B2 lectins.

A, hemagglutinating activity
Purification of B4 Lectin-Because the 1 m N-acetylgalactosamine eluate contained some A, erythrocyte agglutinating activity and trace amounts of band A on SDS-PAGE, the B4 lectin was purified further by anion exchange chromatography. The 1 mM N-acetylgalactosamine eluate (part I) was dialyzed extensively against 10 m Tris-HC1 buffer, pH 8.0, and applied to a DEAE-cellulose column equilibrated in the same buffer. Bound protein was displaced from the column by stepwise elution with NaCl buffered with 10 m Tris-HC1 pH 8.0. As shown in Fig. 4 equilibrium analysis of the lectin displaced with 85 mM NaCl gave a linear plot of log concentration versus radius', indicative of physical homogeneity. The M , calculated from this analysis is 108,300, using a calculated partial specific volume of 0.702 ml/g. The M , of the dissociated subunits, determined by sedimentation equilibrium analysis in the presence of 6 M guanidine HC1, is 25,600. From these analyses, it was apparent that the intact lectin is composed of 4 B subunits, and it has been designated B4.
The 1 mM N-acetylgalactosamine eluates from the preparation summarized in Table I contain approximately 26 mg of B4 lectin, and additional B4 lectin can be purified from the run through material. The purified B4 lectin does not agglutinate A,, B, or 0 erythrocytes at a concentration of c1.15 mg/ml. The protein displaced with 170 m NaCl accounted for 87% of the A, erythrocyte agglutinating units recovered and the SDS-PAGE of this fraction showed that both A and B subunits were present (Fig. 4, inset). The small amount of protein displaced with 500 m~ NaCl was not characterized further.
Purification 0fA4 Lectin-The 100 mM N-acetylgalactosamine eluate was dialyzed extensively against 10 m Na phosphate buffer, pH 7.2, and applied to a DEAE-cellulose column equilibrated in the same buffer. Bound protein was displaced from the column by elution with a linear gradient from 0-0.5 M NaCl buffered with 10 m Na phosphate, pH 7.2. The protein in the 100 mM N-acetylgalactosamine eluate was displaced in 2 peaks with NaCl, as shown in Fig. 5A. Fractions from the fmt peak were pooled as indicated and concentrated.
The inset shows the SDS-PAGE profiles of the 100 mM Nacetylgalactosamine eluate and this concentrated pool, labeled peak 1. SDS-PAGE of peak 1 resolves a single band, with M , = 33,600, i.e. the A subunit, and its electrophoretic mobility does not vary when samples are prepared with P-mercaptoethanol. Sedimentation equilibrium analysis of this lectin in aqueous solution gave a linear plot of log concentration versus radius'. The M, calculated from this analysis is 109,500, using a calculated partial specific volume of 0.722 ml/g. On the basis of the molecular weight of the intact lectins and the estimated molecular weight of the subunit by SDS-PAGE, it appeared likely that this lectin is also composed of 4 subunits, and it has been designated A4.
Approximately 200 pg of & lectin were purified from 100 g of seeds. The purified kc lectin has an A, erythrocyte agglutinating specific activity of 5000 hemagglutinating units/mg.
Purification of A2B2 Lectin-The 10 mM N-acetylgalactosamine eluates were dialyzed extensively against 10 mM Na phosphate buffer, pH 7.2, containing 0.05 M NaCl and applied to a DEAE-cellulose column equilibrated in 10 m Na phosphate buffer, pH 7.2. Bound protein was displaced from the column by elution with a linear gradient from 0.05-0.5 M NaCl buffered with 10 m Na phosphate, pH 7.2. The protein in the 10 m~ N-acetylgalactosamine eluates was displaced in two peaks with NaCl, as shown in Fig. 5B. Fractions from the second peak were pooled as indicated and concentrated. The SDS-PAGE profile of this concentrated pool, labeled peak 2, is shown in the inset. Both A and B bands are present and of approximately equal prominence. Sedimentation equilibrium analysis of this lectin in aqueous solution gave a linear plot of log concentration versus radius'. Since the amino acid and carbohydrate composition of this lectin has not been determined, the partial specific volume was estimated from those of the purified A, and B4 lectins. The M , calculated from this analysis is approximately 94,000.
Chemical Composition of B4 a n d A4 Lectins-The amino acid and carbohydrate analyses of purified B4 and A, lectins are shown in Table 11. Both lectins are rich in the acidic amino   Table I. Column fractions were monitored by absorbance at 280 nm ( -) and by conductivity (---) using a Radiometer conductivity meter. Ears indicate fractions which were pooled and concentrated. A , the 100 m N-acetylgalactosamine eluate was dialyzed against 10 mM Na phosphate buffer, pH 7.2.21.5 ml(0.02 mg of protein/ml) were applied to a DEAE-cellulose column (0.7 X 9 cm) equilibrated in the same buffer and the column was washed with 5 ml of starting buffer. No protein ran through the column and this portion of the profile is not shown. The column was then eluted with a linear 64-ml gradient from 0-0.5 M NaCl buffered with 10 mM Na phosphate, pH 7.2, at a flow rate of 25 ml/h. 1.2-ml fractions were collected. To concentrate peak 1, pooled fractions were diluted to a conductivity of 3.5 m h o with 10 m Na phosphate buffer, pH 7.2, and applied to a DEAE-cellulose column (0.7 X 4 cm) equilibrated in 10 mM Na phosphate buffer, pH 7.2. Bound protein was eluted with 200 mM NaCl buffered with 10 m Na phosphate, pH 7.2. E , the 10 m Nacetylgalactosamine eluates were combined and dialyzed against 10 mM Na phosphate buffer, pH 7.2, containing 0.05 M NaCl. 38.5 ml (0.05 mg of protein/ml) were applied to a DEAE-cellulose column (0.7 X 11.2 cm) equilibrated in 10 mM Na phosphate buffer, pH 7.2, and the column was washed with 5 ml of starting buffer. No protein ran through the column and this portion of the profie is not shown. The column was then eluted with a linear 8 0 4 gradient from 0.05-0.5 M NaCl buffered with 10 mM Na phosphate, pH 7.2, at a flow rate of 32 ml/h. 1.2 ml-fractions were collected. To concentrate peak 2, pooled fractions were diluted to a conductivity of 3.5 mmho with 10 m Na phosphate buffer, pH 7.2, and applied to a DEAE-cellulose column (0.7 X 5.7 cm) equilibrated in 10 mM Na phosphate buffer, pH 7.2. After eluting the column with 100 m NaCl, bound protein was eluted with 250 mM NaCl buffered with 10 m Na phosphate, pH 7.2.
The inset shows the SDS-PAGE profiles of a concentrated 100 mM N-acetylgalactosamine eluate and peaks 1 and 2, concentrated as described above. 5-10 pg of each were applied to the gel. Arrows mark the position of the molecular weight standards. acids, aspartic acid and glutamic acid, and the hydroxylic amino acids, serine and threonine. In addition, neither lectin contains any cysteine or methionine. The B4 lectin contains mannose, N-acetylglucosamine, and fucose, and these comprise 9.8% of its weight. The & lectin, which is composed of 6.7% carbohydrate by weight, contains galactose in addition to those sugars. Glucose was also present in the carbohydrate analyses of the purified lectins but because its content varied considerably among analyses, it was assumed to be a comtaminant. A monosaccharide having a retention time identical to that of xylose was also present, but this was not analyzed further.  Purified Lectins to A1 Erythrocytes-Fig. 6 shows the binding of purified lectins to A1 erythrocytes, performed as described under "Experimental Procedures." The curve demonstrates that binding of the & lectin to A1 erythrocytes at 4 "C is saturable. The Scatchard plot of these data, shown in the inset, is linear. The association constant derived from this analysis is 1.8 X lo7 M" (Kd = 5.5 X lo-' M) and

Binding of
there are approximately 273,000 binding sites/cell. SDS-PAGE and autoradiography of the erythrocyte pellet confirmed that the bound radiolabel was associated with intact A subunits (not shown). In contrast, B4 lectin does not bind to Al erythrocytes. The AzBz lectin also bound to A1 erythrocytes, with a similar number of binding sites per cell and an association constant approximately half of that of the & lectin (data not shown). SDS-PAGE and autoradiography of the erythrocyte pellet showed that bound radiolabel was associated with both A and B subunits (not shown). Since the B4 lectin does not bind to A1 erythrocytes, finding radiolabel associated with the B subunit in the erythrocyte pellet provides additional evidence for the presence of a lectin composed of both A and B subunits.

DISCUSSION
The data presented in this paper demonstrate that V. villosa seeds contain at least three different lectins composed of two different subunits. The relationship of these lectins to the lectin preparations in the literature is not entirely clear .  Kimura et al. (3) prepared the lectin used in their initial lymphocyte binding studies by affinity chromatography on immobilized human blood group A substance. Subsequently, others have prepared lectin using an affinity adsorbent of Nacetylgalactosamine coupled to epoxy-activated Sepharose 6B (5, 12). We have used a column of immobilized porcine blood group substances. The glycopeptides found in porcine stomach mucin are heterogeneous (26), although some of the structures have A blood group reactivity.
We first attempted to prepare V. vitlosa lectins according to the published method (3) in which bound material was eluted from the affinity column with 10 m~ N-acetylgalactosamine. Analysis of fractions in this eluate revealed 1) that the Al erythrocyte agglutinating specific activity increased markedly across the protein peak and 2) that 2 bands of protein were resolved on SDS-PAGE. Furthermore, the lower of the 2 bands became more prominent across the protein peak, in parallel with the increasing AI erythrocyte agglutinating specific activity. This observation suggested that the hemagglutinating activity was contaminated with another protein that bound less tightly to the affinity column, and we subsequently isolated lectins by elution with increasing concentrations of N-acetylgalactosamine.
The yield of lectins prepared by chromatography on various adsorbents is quite similar, i.e. 75-80 mg/100 g of seeds. Our work indicates that most of the protein that binds to the affinity column (86% in the preparation summarized in Table   I) is eluted with 1 nm N-acetylgalactosamine and that the most prevalent lectin in this eluate, i.e. the B4 lectin, has no A1 erythrocyte agglutinating activity. The 5.2 mg (the remaining 14%) of protein eluted from the affinity column with >1 mM N-acetylgalactosamine accounted for 54% of the hemagglutinating activity recovered. Because the eluates contain mixtures of lectins and because all of the hemagglutinating activity in the seed extract was not recovered, it is difficult to estimate from these data the amount of A1 erythrocyte agglutinating lectins present in the seed extract. On the other hand, if we assume that the A subunit-containing lectins have a specific activity of 2500-5000 hemagglutinating units/mg, 100 g of seeds could contain no more than 17.8 mg of these lectins.
Although the AI erythrocyte agglutinating specific activity is not reported in other published lectin preparations, our data suggest that these preparations probably do not contain exclusively the AI erythrocyte agglutinating lectins. The report of Grubhoffer et al. (27) suggested that the Al erythrocyte agglutinating activity of different cultivars within the species V. villosa varied. These workers prepared lectin from V. uillosa seeds using an affinity column of N-(~-ami-nohexanoy1)-D-galactosamine coupled to Sepharose 4B. The yield of a purified lectin prepared was 22 mg/100 g of seeds, and this lectin agglutinated human blood group A, erythrocytes at a minimum concentration of 15 pg/ml. Their hemagglutination inhibition studies differ significantly from those reported in the accompanying paper for the B4 lectin, suggesting that they have in fact characterized the AI erythrocyte The B, lectin was readily purified from the material eluted with 1 mM N-acetylgalactosamine from the affinity column, since it was by far the most prevalent lectin in this eluate.
Purification of the & lectin was less straightforward. Most (81%) of the hemagglutinating activity that remained bound to the affinity column in the presence of 1 m~ N-acetylgalactosamine was eluted with S10 mM N-acetylgalactosamine. As described under "Results" and shown in Fig. 3, the upper of the 2 major bands (band 2) in the electrophoretic profile of the 10 m~ N-acetylgalactosamine eluate contained both A and B subunits, although A was more prominent. Since gel electrophoresis of the B4 lectin under nondenaturing conditions revealed that it migrated as a rather broad band, it is likely that band 2 in fact represents the & lectin and a contaminating amount of Bd lectin. Nevertheless, we have been unable to separate completely the & lectin from the B4 lectin in the 10 nm N-acetylgalactosamine eluate by anion exchange chromatography. SDS-PAGE analysis of column fractions from anion exchange chromatography of the 10 mM N-acetylgalactosamine eluates (Fig. 5 B ) revealed that both A and B subunits were present in the first peak eluted (data not shown). As shown in Fig. 3, the upper of the 2 major bands in the electrophoretic profile of the 100 m~ N-acetylgalactosamine eluate contained only A subunit, and we subsequently purified the A4 lectin from this eluate by anion exchange chromatography. As shown in Fig. 3, the lower of the 2 major bands in the electrophoretic profiles of both the 10 and 100 mM N-acetylgalactosamine eluates contained both A and B subunits, of approximately equal prominence. We purified the AzBz lectin from the 10 nm eluate because it was present there in a somewhat larger amount and because the 100 mM eluate also contained a protein which migrated just below the lower major band on gel electrophoresis.
The purified B4 and A4 lectins have similar molecular weights, and analysis of the lectins under denaturing conditions suggests that each is composed of four apparently identical subunits. The B subunit appears to have a somewhat larger apparent molecular weight by SDS-PAGE, perhaps because it is more heavily glycosylated. The chemical compositions of the purified lectins are remarkably similar and characteristic of plant lectins generally (28). The B4 lectin does not bind to A1 erythrocytes and does not agglutinate A, B, or 0 erythrocytes. In contrast, there are approximately 273,000 binding sites for the & lectin on AI erythrocytes. The number of binding sites per cell is similar to those reported using other blood group AI agglutinins and to the receptor number determined with the blood group B-specific lectin from Griffonia simplicifolia (29).
The similarity of the amino acid composition of the B and A subunits and the identification of at least one lectin which appears to be composed of both A and B subunits suggest that agglutinin.

Lectins from V.
V. villosa seeds may contain a family of isolectins, i.e. tetrameric structures composed of two unique subunits in various proportions. Such families of isolectins are not uncommon, the best characterized being those of G. simplicifolia I (30, 31) and Phaseolus vulgaris (32, 33). The subunits in both of these are indistinguishable in immunochemical reactivity and have similar amino acid compositions, but, interestingly, differ in their carbohydrate binding specificity. The G. simplicifolia I A, isolectin agglutinates A but not B erythrocytes, whereas G. simplicifolia I AB3 and B4 lectins are highly blood group B-specific. Within the lectins derived from P. vulgaris, the L subunit is a potent leukoagglutinin and mitogen that lacks erythroagglutinating properties whereas the E subunit is a potent erythroagglutinin with little or no mitogenic activity. Their carbohydrate binding specificities have recently been elucidated (34, 35).
It is not clear which of the lectins we have described interacts with the oligosaccharide structure of T145 on cytotoxic T lymphocytes. Preliminary binding studies in our laboratory using purified B4 and & lectins and the cloned lines