T., Matsumoto, I., and Osawa, T. (1974) J. Biol. Chem. 249, 2786). We now report that leukoagglutination

Transcription

1 THE JOURNAL OF BOLOC~CAL CHEMSTRY 0 19% hy The American Society for Biochemistry and Molecuter Biology, nc Val. 263, No. 10, issue of April 5, pp , 1988 Printed in U.S.A. The mmobilized Leukoagglutinin from the Seeds of Mauckia amurensis Binds with High Affinity to Complex-type Asn-linked Oligosaccharides Containing Terminal Sialic Acid-linked a-2,3 to Penultimate Galactose Residues* (Received for publication, August 18,1987) Wei-Chun Wang and Richard D. CummingsS From the Department of Biochemitry, School of Chemical Sciences, University of Georgia, Athens, Georgia We recently reported that the purified leukoaggfutinin (designated MAL) from the seeds of the leguminous plant Maackia amurewis is a potent leukoagglutinin for the mouse lymphoma cell line BW6147 (Wang, W.-C., and Cummings, R. D. (1987) Anal. Biochem. 161,80). We and others have shown that this lectin is a weak ~emagglutin~n of human erythrocytes (Kawaguchi, T., Matsumoto,., and Osawa, T. (1974) J. Biol. Chem. 249, 2786). We now report that leukoagglutination by MAL is inhibited by low concentrations of 2,3-sialyllactose (NeuAcaB,3Galf31,4Glc), but it is not inhibited by either 2,6-sialyllactose (NeuAca2,6GalB- 1,4Glc), lactose, or free NeuAc. To further study the carbohydrate-bin~ng specificity of this lectin, we investigated the interactions of immobilized MAL with glycopeptides prepared from the mouse lymphoma cell line BW5147 and from purified glycoproteins. We found that immobilized MAL interacts with high affinity with complex-type tri- and tetraantennary Asn-linked oligosaccharides containing outer sialic acid residues linked a2,3 to penultimate galactose residues. Glycopeptides containing sialic acid linked only inhibited by Siaa2,3Gal@1,4Glc, but it is not inhibited by a2,6 to penultimate galactose did not interact detecta- Siaa2,6Gal@l,4Glc. These observations led us to investigate bly with the immobilized lectin. Our analyses indicate the possibility that immobilized MAL might be useful in thathe interactions of complex-type Asn-linked separating mixtures of sialylated oligosaccharides containing chains with the lectin are dependent on sialic acid sialic acid in either a2,3- or a2,6-linkage to galactose. We linkages and are not dependent on either the branching have found that the immobilized MAL interacts with high pattern of the mannose residues or the presence of affinity with complex-type Asn-linked oligosaccharides conpo~y-~-acety~lact~mine sequences. taining terminal sialic acid in a2,3-linkage to galactose. We demonstrate both the interaction of MAL with glycoconjugates containing sialic acid in a2,3-linkage and the utility of the immobi~ized lectin in the separation and fractionation of Plant lectins have been used to determine the location of complex mixtures of sialylated oligosaccharides. glycoconjugates in animal cells and to isolate and purify glycoconjugates (1-6). Such applications have expanded in EXPER~ENTAL PROCEDURES recent years since several lectins have been found to interact with high affinity with specific determinants of animal cell glycoconjugates. However, the use of lectins in these types of studies is limited because the oligosaccharide-bin~ng speci- ficity of relatively few plant lectins has been studied in detail. Furthermore, there are many animal cell glycoconjugates * This work was supported by the following research grants (to R. D.C.): Research Grant RO-1 CA37626 from the National Cancer Xnstitute, a Faculty Research Grant from the University of Georgia Research Foundation, and Biomedicai Research Support Grant SO7 RR from the National nstitutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked vertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. $ To whom correspondence should be addressed which are not known to interact specifically with any known plant lectin. For example, in many glycoconjugates sialic acid is a major terminal sugar (7-10) occurring as a number of derivatives (11) in a variety of linkages to other sugars (7-10). Some of the commonly found sialylated sequences are Siaa2,3- Gal@l,~GlcNAc-R, Siaa2,6G~~l,4GlcNAc-R, Siaa2,3Gal@l,3GlcNAc-R, and Siaa2,6Gal- NAcR (7-10). However, only one plant lectin has been found to efficiently discriminate among these types of sialylated sequences. The hemagglutinating lectin from eiderberry bark (Sambucus nigra L.) binds with high affinity to glycoconjugates containing the a2,6-linked sialic acid, while isomeric structures containing terminal sialic acid in a2,3-linkage are not bound (12). The immobilized elderberry bark lectin has been used to separate glycoconjugates containing a2,6-linked sialic acid from those lacking this terminal sequence (13). During our investigations into the carbohydrate-binding specificity of a le~oagglutinin MAL from the seeds of Maackia amurensis (14, S), we found that leukoagglutination of the mouse lymphoma cell line BW5147 by the lectin is Materials-Sephadex G-25, QAE-Sephadex, Dowex 50 (H form), sodium borohydride, porcine thyroglobulin, bovine thyroglobulin, bovine fetuin, cyanogen bromide-activated Sepharose-$B, bovine epididymal a-l-fucosidase, N-acetyl-D-glucosamine, N-acetyl-D-galactosamine, N-acetylneuraminic acid, lactose, D-galactose, D-mannose, D-glucose, and Amberlite MB-3 were obtained from Sigma. D-[2-aH] Mannose (24 Ci/mmol~ and ~-[6-~H]galactose (25 Ci/mmol) were purchased from CN and [l- *C]acetic acid anhydride (111 mci/ mmol) was obtained from Amersham Corp. Arthrobacter ureafmiens neuraminidase was purchased from Boehringer Mannheim and Vibrio choiem neuramini~e and Pronase were obtained from Behring Diagnostics. ConA-Sepharose was obtained from Pharmacia LKB The abbreviations used are: MAL, Maackia arnurensis leukoagglutinin; ConA, concanavalin A; SDS, sodium dodecrl sulfate; HPLC, high performance liquid chromatography; PBS, phosphate-buffered saline; and TBS, Tris-buffered saline.

2 nteractions of Sialylated Glycoconjugates with mmobilized MAL 4577 Biotechnologies nc. L,-PHA-Agarose and pea lectin-sepharose were purchased from E-Y Laboratories, nc. Bio-Gel P-10 was obtained from Bio-Rad. a-mannosidase, 6-N-acetylhexosaminidase, and P- galactosidase were prepared from jack bean meal (16). Silica-gel G- coated thin layer chromatography plates were obtained from Analabs. Tissue culture media was purchased from Flow Laboratories. Escherichia freundii was the kind gift of Dr. Minoru Fukuda (La Jolla Cancer Research Foundation, La Jolla, CA). The oligosaccharides and were kindly supplied by Dr. David Smith (Virginia Polytechnic nstitute, Blacksburg, VA). The following standard oligosaccharides were prepared from BW5147 cell-derived glycopeptides as described previously (17): and M. amurensis seeds were obtained from The Arnold Arboretum of Harvard University (Jamaica Plain, MA). Purification of MAL and Preparation of MAL-Sepharose-MAL was purified from M. amurensis seeds by a slight modification of the procedure of Kawaguchi et al. (15), as discussed below. From 25 g of seeds we obtained 171 mgof the leukoagglutinin, which we have designated MAL, for M. amurensis leukoagglutinin. (n the published procedure for purifying this lectin (15), the authors refer to the lectin as MAM, a designation which denotes the mitogenic activity of the lectin.) Purification of MAL was monitored by a leukoagglutination assay using the mouse lymphoma cell line BW5147 (see below). Kawaguchi et al. (15) partially purified MAL by affinity chromatography over a column of Sepharose-4b containing covalently bound glycopeptides derived from porcine thyroglobulin. nstead, we prepared and utilized a column (1.5 X 19 cm) containing desialylated intact porcine thyroglobulin attached to CNBr-activated Sepharose- 4B. The glycoprotein was coupled at a density of 12 mg/ml of gel. The MAL in crude extracts bound to this support and was eluted by applying 100 mm of glycine buffer, ph 3. The eluted lectin was dialyzed against phosphate buffer at ph 4.5 and applied to a column of SP-Sephadex at the same ph. MAL did not bind to the resin at this ph and by this procedure was separated from contaminating levels of the hemagglutinating lectin, as shown previously (14). The hemagglutinating activity was bound completely by the resin. The punty of the MAL was assessed by polyacrylamide slab gel electrophoresis in 10% acrylamide in the presence of SDS and 2- mercaptoethanol (18). Upon staining with Coomassie Blue, we observed two bands of nearly equal intensity with apparent Ma of 38,000 and 40,000. These results are consistent with those of Kawa- guchi et al. (15) in which they observed that the purified MAL migrated as a rather broad band with apparent M, of 35,000 upon electrophoresis in polyacrylamide tube gels in SDS and 2-mercaptoethanol. Purified MAL (15 mg) was suspended in 2 ml of coupling buffer containing 0.2 M NaHC03, 50 mm GalNAc, ph 7.9. To thisuspension a moist cake of CNBr-activated Sepharose-4B was added (prepared from 1.5 g of dry gel), which had been washed previously with icecold 0.2 M NaHCO,, ph 7.9 (20 ml). The suspension was mixed gently for 24 h at 4 "C, after which the Sepharose was allowed to settle and the supernatant containing uncoupled MAL was removed. The amount ofmal recovered in the supernatant was determined to estimate the amount of lectin coupled. Protein determinations were performed by the method of Lowry et al. (19), using bovine serum albumin as a standard. Coupling efficiency was estimated to be approximately 80%, and the lectin was coupled at a density of 2.5 mg of MAL/ml ofgel. The MAL-Sepharose was resuspended in the coupling buffer with 1 M ethanolamine, ph 7.9, and allowed to incubate for 2 h at 4 C. The MAL-Sepharose was then transferred to a column (0.6 X 17 cm) and washed with phosphate-buffered saline containing 6.7 mm KH2P04, 150 mm NaCl, 0.02% NaN3, ph 7.4 (PBS-NaN3). The column of MAL-Sepharose was stored routinely in this buffer at 4 C. No appreciable loss of activity was observed for the column over a period of 6 months. Leukoogglutinatwn Assays-Leukoagglutination assays were performed as described by Wang and Cummings (14), using BW5147 mouse lymphoma cells stained with neutral red. Briefly, the cells were stained for 30 s at a concentration of 5 X O6 cells/ml in PBS, containing 0.1% neutral red (w/v). The stained cells were collected by centrifugation and resuspended in PBS at a concentration of 5 X 10' cells/ml. The stained cells (15 pl) were placed on a glass slide and mixed with 10 pl of PBS and 10 pl of a solution of MAL in PBS. The sample was stirred gently with a glass rod for 3 min at room temperature. Agglutination was then scored visually on a scale of 0 to 4+ with 0 representing no agglutination and 4+ representing complete agglutination of all visible cells. To assess the inhibition of leukoagglutination by purified oligosaccharides and sugars, standard solutions of these individual com- pounds (shown in Table ) were prepared in PBS and various concentrations of the compounds were included in the leukoagglutination assay in place of the 10 pl of PBS. For these inhibition assays we used the lowest concentration ofmal (a 1% titer of a 1 mg/ml solution) which gave a 4+ agglutination reaction. The concentration of potential inhibitor reducing the agglutination from 4+ to a 1-2+ was taken as the concentration necessary for 50% inhibition of leukoagglutination under these conditions. Lectin Affinity Chromatography of Glycopeptides-Chromatography of glycopeptides on ConA-Sepharose, pea lectin-sepharose and L4-PHA-Agarose was performed exactly as described previously (2, 20). For chromatography on MAL-Sepharose, glycopeptides were dissolved in 0.5 ml of PBS-NaN3 and applied to the washed column. Either 1 ml or 0.5-ml fractions were collected, as indicated in the figures, and portions of the fractions wereremoved to determine radioactivity by liquid scintillation counting. For the MAL-Sepharose columns used all glycopeptides and standards not bound by the lectin (noninteractive material) eluted in a volume of approximately 4.5 ml. All chromatographic procedures were performed at room temperature. Preparation of Metabolically Radiolabeled Glycopeptides from the Mouse Lymphoma Cell Line BW5147-The mouse lymphoma cell line (BW5147) was cultured in the a modification of Eagle's minimal essential media containing 10% horse donor serum. To metabolically radiolabeled glycoprotein oligosaccharides in these cells, 1.5 X 10' cells were incubated in culture for 48 h with D-[2-3H]mannose or D- [6-3H]galactose(0.05 mci/ml) in 30 ml of complete culture media. Glycopeptides were prepared by treatment of lysed cells with Pronase and radiolabeled glycopeptides were separated by serial lectin affinity chromatography, as described previously (17, 20,21). Glycopeptides from the cells were designated as A1, A2, B1, B2, and A, as described previously (20). The partial structures of the Asn-linked oligosaccharides in these glycopeptides have been described previously (17, 20, 21) and are discussed under "Results." Further characterizations of many of these glycopeptides are provided in this study. A1 are glycopeptides not bound by ConA-Sepharose, pea lectin-sepharose, and L,-PHA-Agarose. A2 are glycopeptides not bound by either ConA-Sepharose or pea lectin-sepharose but which are bound by L4-PHA-Agarose. B1 are glycopeptides not bound by either ConA-Sepharose or L4-PHA-Agarose but which are bound by pea lectin-sepharose. B2 are glycopeptides not bound by ConA-Sepharose but which are bound by both pea lectin-sepharose and L4-PHA-Agarose. A are glycopeptides bound by ConA-Sepharose and eluted with 10 mm a-methylmannoside but which are not bound by pea lectin-sepharose. lsolution of Standard Glycopeptides from Thyroglobulin and Fe- tuin-glycopeptides containing the complex-type Asn-linked oligosaccharides from bovine fetuin and bovine and porcine thyroglobulin were prepared by extensive treatment of the glycoproteins with Pronase. Briefly, 1 g of each of the glycoproteins was treated with 30 mg of Pronase in a final volume of 6 ml for 24 h at 60 "C in 0.1 M Tris- HCl, 1 mm CaC12, ph 8.0. The treated material was desalted by passage over a column of Sephadex G-25 (1 X 60 cm) in 7% n-propyl alcohol in water. The glycopeptides containing the N-linked sugar chains eluted in the void volume and were pooled and dried. Of the three glycoproteins used in this study, only fetuin is known to contain 0-linked sugar chains which consist primarily of tri- and tetrasaccharides (22). Under these conditions, the glycopeptides containing 0-linked chains from fetuin are included in the Sephadex G-25 column. For further purification the large-sized glycopeptides were resuspended in Tris-buffered saline (TBS), consisting of 10 mm Tris-HC1, 150 mm NaCl, 1 mm CaC12, 1 mm MnC12, and 0.02% NaN3, ph 8.0, and were applied to a column of ConA-Sepharose (60 ml) equilibrated in TBS. Five-ml fractions were collected and unbound glycopeptides were eluted with TBS (250 ml), whereas bound glycopeptides were eluted with 10 mm a-methyl-glucoside in TBS (250 ml). Some of the glycopeptides from porcine and bovine thyroglobulin contain high mannose-type N-linked chains (23, 24). These glycopeptides are bound with high affinity by ConA-Sepharose and their elution from the immobilized lectin requires 100 mm a-methyl-mannoside. Fractions containing unbound glycopeptides and those eluted with a- methyl-glucoside were pooled separately and dried in a rotary flash evaporator under reduced pressure. The samples were resuspended in

3 4578 nteractions of ~ialy~ted ~~ycoconj~a~es with ~ r n ~ ~ MAL b i ~ ~ ~ e ~ a small volume of water. Salt and hapten sugars were separated from glycopeptides by chromatography on a column of Sephadex G-25, as described above. Glycopeptides were stored routinely in water at -20 "C. Glycopeptides were quantified after determining the content of neutral hexose by the phenol/sulfuric acid assay (25). The structures of the major species of glycopeptides isolated from the various glycoproteins are shown in Fig. 1. Previous studies have shown that fetuin contains a heterogeneous array of sialylated complex-type triantennary N-linked chains (26-28). The complex-type N-linked chains from fetuin are not bound by ConA-Sepharose, and they were used directly in our studies. The structure of one of the major species from fetuin is shown in Fig. 1 as glycopeptide (26-28). Previous studies have shown that many of the glycopeptides from bovine thyroglobulin are not bound by ConA-Sepharose and are mixtures of compiex-type tri- and tetraantennary N-linked chains with the structures depicted in Fig. 1 as glycopeptides, V, and V (24). For our studies glycopeptides ZZ, ZV, and V were not separated from each other and were utilized as mixtures. The major glycopeptide from bovine thyroglobulin that is bound by ConA-Sepharose and eluted with a-methyl-glucoside is a complex-type biantennary N- linked chain with the structure of glyco~ptide ZZ (24). Porcine thyroglobulin contains both complex-type bi- and triantennary N- linked chains (23). The biantennary chains from porcine thyroglobulin, having the structure of glycopeptide V (Fig. l), are bound by ConA-Sepharose and eluted with a-methyl-glucoside, while the triantennary species, having the structure of glycopeptide V (Fig. l), are not bound by the immobilized lectin (23). Radiolabeling of Glycopeptides by N-Acetylation Using "C-Labeled Acetic Anhydr~-Glycopeptides through V1 (Fig. 1) were radiolabeled by 1-%-N-acetylation, as described by Finne and Krusius (29). The glycopeptides (10 nmol) were dissolved in 0.1 ml of water and 0.2 mi of 1 hi NaHC03 was added. After [1-"Clacetic anhydride (10 pci) was added, the mixture was incubated at room temperature for 30 min. Then 0.2 ml of 1 M NaHCO, was added, followed by 0.2 ml of a freshly prepared 2% solution of unlabeled acetic anhydride in acetone. The mixture was acidified by 0.2 ml of 4 N acetic acid. Free radioactivity was removed by repeated evaporation from 0.1 M pyridinelacetic acid buffer, ph 5.0, and by subsequent chromatography on a column of Sephadex G-25 in 7% n-propyl alcohol. Approximately 1 X 10' cpm of l-'%-acetylated glycopeptides were recovered from each of the labeled samples. Column Chromatography of Glycopeptides on QAE-Sephader-Col- umn chromato~aphy on Q~-Sephadex was performed as described by Varki and Kornfeld (30). Briefly, a 2-ml column of QAE-Sephadex was prepared in 2 mm Tris base, ph 9.0. Glycopeptides were applied to the column and 2-ml fractions were collected. Bound glycopeptides were eluted in with 2 mm Tris, ph 9.0, containing increasing concentrations of NaC1. Portions of the collected fractions were removed, and radioactivity was determined by liquid scintillation counting. Methylation Analysis-[3H]Galactose-laheled glycopeptides were methylated by the procedure of Hakomori (31) and then hydrolyzed as described previously (21, 32). The methylated galactose species were separated by thin layer chromatography on Silica-gel G in the solvent system acetone/wa~r/~monium hydroxide ( ) (33). [3H]Mannose-labeled glycopeptides were also methylated, as described above, and the methylated mannose and fucose residues were separated directly by either thin layer chroma~graphy or by high performance liquid chromatography after reduction of the methylated sugars with NaBH4. Thin layer chromatography of the methylated mannose and fucose residues was conducted in acetone~nzene/ water/ammonium hydroxide (80: ). The sample lanes were scraped in 0.5-cm sections and radioactivity was determined by scintillation counting. Methylated standards were obtained as described previously (20,21). For HPLC analysis, the methylated [3H]mannoselabeled glycopeptides were hydrolyzed, and the resulting methylated mannose and fucose residues were reduced with NaBH, and separated by reverse-phase HPLC (Zorbax ODS; 250 X 4.6 mm) at 45 "C under the conditions described by Sziagyi et ul. (34). HPLC Analysis of PH/Galuctose-tabeled Ofigosaccharides Released from Glycopeptides by Endo-@-galactosidase-Glycopeptides were treated with E. frettndii endo-~-galactosid~e using the conditions described previously (17, 21). [3H]Galactose-labeled di- tri-, and tetrasaccharides released from glycopeptides by endo-&galactosidase were separated by HPLC on a Varian Micropak AX-5 column (4 X 300 mm) at room temperature using a Beckman lloa HPLC pump and 420 controller. The elution program was 20% water with 80% acetonitrile from time 0 to 2 min; from time 2-42 min a linear gradient of water from 20 to 60% was used. The flow rate was 1 ml/min and 0.5-ml fractions were collected. Paper Chromatography of Oligosaccharides-Oligosaccharides were separated by descending paper chromatography on Whatman No. 1 paper in ethyl acetate/pyridine/acetic acid/water (551:3). Each lane was cut into 1-cm sections, and radioactivity in each section was determined by liquid in till at ion counting. n preparative paper chromatography each of the sections was soaked in 1 ml of water and a portion of sample was removed for liquid scintillation counting. ~es~lylation and Enzymetic ~ es~y~tion of Gly~o~ptides-Glycopeptides were desialylated by treatment with 2 N acetic acid at 100 "C for 1 h. The acid was removed by evaporation under reduced pressure. Glycopeptides were also desia~y~ated enz~atica~ly using A. ureafaciem sialidase. n this case glycopeptides were suspended in 100 mm sodium acetate, ph 4.8 (50 pl) containing 10 milliunits of the enzyme and incubated for 12 h at 37 "C. n control incubations, the enzyme was boiled for 10 min in the presence of 50% ethanol to insure complete denaturation of the enzyme. Ethanol was removed by evaporation under reduced pressure, and the treated enzyme was resuspended in water and added to the reaction mixtures. Sialic acid was added to desialylated glycopeptides by Dr. James Paulson (UCLA School of Medicine) by incubation of the glycopep- tides with either purified CMPNeuAc:Gal@(1,3/4)GlcNAc a2,3-sialyltransferase from rat liver (35, 36) or purified CMPNeuAc: Gal@l,.tGlcNAc a2,6-sialyltransferase from rat liver (35, 36). Before enzymatic resialylation, the glycopeptides were desialylated by treatment with mild acid, as described above. The desialylated glycopeptides were then incubated overnight with 18 nmol (3312 cpm/nmol) CMP-('4C]sialic acid and 2 milliunits of either a2,3- or a2,6-sialyltransferase, as described (35, 36), in a final volume of 90 pl for a2.3- sialyltransferase or 78 pl for a2,6-sialyltransferase. After treatment, the CMPNeuAc was separated from glycopeptides by passage over a column of Sephadex G-25 (1 X 60 em) in 0.1 M pyridine/acetate, ph 5.0. The glycopeptides were recovered in the void volume and were dried by evaporation under reduced pressure. RESULTS nhibition of MAL-dependent Leubagglutination by Purified Oligosaccharides and Momsaccharides--n our initial studies on the carbohydrate-binding specifieity of MAL, we investigated the inhibition of leukoagglutination of BW5147 cells by several purified oligosac~harides and monosaccharides. As shown in Table, we found that leukoagglutination was detectably inhibited by only one of the compounds tested, the trisaccharide Siaa2,3GalB1,4Glc, in which case 50% inhibition was observed at a concentration of 2 mm. nterestingly, neither the isomeric sialylated ol~gosacch~i~e Siaa2,6Galpl,4Glc nor free N-acetylneuraminic acid were effective inhibitors of leukoagglutination. These results strongly suggested that MAL binds with high affinity to sialylated glycoconjugates containing terminal sialic acid in &,3-linkage to penultimate galactose. Affinity Chromatography of Purified Glycopeptides on MAL- Sepharose--En view of the above findings, we investigated the carbohydrate determinants required for high affinity binding of several purified glycopeptides to MAL. MAL was immobi- TABLE 1 nhibition of~a~-i~~ed leu~~~ut~~tion of BW5147 cells by bw molecular weight sugars Cone. required for 50% Sugars tested inhibition of leukoagglutination NeuAca2,3Gal@l,4Glc NeuAc~~2,6Gal@l,4Glc Lactose N-Acetylneuraminic acid o-galactose D-GlUCOSe N-Acetyl-D-giucosamine N-Acetyl-D-galactosamine D-Mannose 2 mm No inhibition at 80 mm

4 nteractions of Sialylated Glycoconjugates with mmobilized MAL 4519 lized to Sepharose 4B at a density of 2.5 mg/ml gel, and a small column (0.6 X 17 cm) of MAL-Sepharose in PBS-NaN, was prepared. Purified glycopeptides of known structures obtained from bovine and porcine thyroglobulin and bovine fetuin were radiolabeled by N-acetylation with [ 14C]acetic anhydride. The glycopeptides were analyzed for their interactions with MAL- Sepharose. n Fig. 1 are shown both the structures of standard glycopeptides tested and the results of whether the glycopep- tides interacted with the immobilized MAL. n Fig. 2, a and b, we have shown representative elution profiles of glycopeptides bound and not bound by MAL-Sepharose. Of all glycopeptides tested, only glycopeptide, which was derived from bovine fetuin, demonstrably interacted with the immobilized MAL; the glycopeptide was retarded in its elution from the column (Fig. 2b). Under these conditions, elution of glycopeptide did not require haptenic sugars or any special elution buffer. Approximately 65% of the fetuin-derived glycopeptide GLYCOPEPTDE DESGNATON GLYCOPEPTDE STRUCTURE ELUTON RETARDED BY HAL-SEPEAROSE 1 11 NcuNAca2-3Gal01-4GlcNAc~l-P.aal. gkn~l-4glcnac~l-4glcnac9l-asn NeuNAca2-6Gal01-4GlcNAc01-2Uanal 4 NcuNAca2-3Ga101-4GlcNk~l C.l01-4Gl~HAc01-2lluull, - +UeuNAca2-6 ~n01-4glcnac01-bglcnac8l-asn - Gal01-bGlcNAc01-21 Pucal-6 +JeuNAca2-6 Ga181-4GlcNAc01-2Uewrl\6 pn01-4glcnac01-4glcnac01-asn - Ga101-4GlcNAc01-2Uenal - 4 Ga101-4Gl~NA~ (Gal01-4GlcNAc01, 6 V V Gal01-4GlcNAcB1, 6 - +NcuNAca2-6 C.161-4GlcNA~01-2Ua1~ll. 6Man~1-4GlcNAc~1-4GlcNAc01-Aen - C.l~1-4GlcNAc~1-2Hanal~3 4 Ga101-4GlcNAc01 Fucal-6 V V1 +NeuNAca2-6 Gal~1-4Gl~NAc61-21Lnal~ gkn01-4glcnacbl-4glcnac01-aen - +NeuNAca2-6 C.1~1-4GlcNAc~ Pucal-6 1 +NauNAca2-6 C.101-4Glc#Ac ~01-4GlcNA~01-4GlcNAcBl-Aen - McuNAca2-6 Ga1~1-4GlcNAc~1-W1~ - 4 Puca 1-6 Ga1814ClcNAc~l FG. 1. Structures of glycopeptides and their interactions with MAL-Sepharose. The glycopeptides tested are designated Z-VZZ and their structures are shown. On the right side of the figure we have indicated whether or not the glycopeptides were retarded in their elution from the column of MAL-Sepharose (+ indicates retardation and - indicates no retardation). Glycopeptide Z represents one of the major species in glycopeptides derived from bovine fetuin (26-28, 56). Glycopeptides ZZ,ZZZ,ZV, and V are major glycopeptides from bovine thyroglobulin (24). Glycopeptides ZZZ, ZVand Vwere tested as mixtures for their interactions with MAL-Sepharose. Glycopeptides VZ and VZZ are major glycopeptides from porcine thyroglobulin (23). Glycopeptides were radiolabeled by 1- C-N-acetylation and applied to a column of MAL-Sepharose (see Fig. 2).

5 4580 nteractions of Sialyluted Glycoconjugates with mmobilized MAL - k,oo0t 4 1 FRACTON (O.5ml) FG. 2. Chromatography of C-N-acetylated glycopeptides on MAL-Sepharose. a, a mixture of 14C-N-acetylated glycopeptides, V, and V (Fig. 1) was applied to MAL-Sepharose; b, the C-Nacetylated glycopeptide (Fig. 1) was applied to MAL-Sepharose before (. -.) and after (A-A) treatment with mild acid to effect desialylation, as described under Experimental Procedures. Recovery of applied radioactivity in each case was greater than 95%. The elution volume of unbound and noninteractive material was approximately 4.5 ml and was contained in fraction numbers 9 and 10. was retarded in its elution from the column and two major peaks of retarded material were observed (Fig. 26). Desialylation of the glycopeptide by treatment with either mild acid or Arthrobacter neuraminidase abolished the interactions of the glycopeptides with immobilized MAL (Fig. 26). All of the glycopeptides shown in Fig. 1 contain sialylated galactose residues, however, only glycopeptide contains terminal sialic acid residues in a2,3-linkage to galactose. These results are consistent with the hapten inhibition study described above, which indicated that glycoconjugates containing the terminal sequence Siaa2,SGal-R interact with high affinity with MAL. Several groups have reported microheterogeneity of sialylation of the complex N-linked oligosaccharides of fetuin (26-28). Sialic acid is present in both a2,3- and a2,6-linkages to galactose, and it is probable that there are many forms of the sialylated glycopeptides differing in the number of sialic acid residues and in the proportion of a2,3-linked and a2,6-linked sialic acid (26-28). This heterogeneity may explain the apparent fractionation of the fetuin-derived glycopeptide on MAL-Sepharose (Fig. 26). t is probable that those glycopeptides most retarded contain more cu2,3-linked sialic acid than those less retarded by the immobilized MAL. solation of Glycopeptides from B W5147 Cells That nteract with High Affinity with MAL-Sephurose-To further investigate the carbohydrate-binding specificity of MAL and the utility of MAL-Sepharose for fractionating complex mixtures of glycopeptides, we examined the interaction of cell-derived glycopeptides with MAL-Sepharose. Since BW5147 cells are agglutinated by MAL, we isolated the glycopeptides from this cell line, fractionated them by affinity chromatography on b MAL-Sepharose, and determined the sialylation pattern of the interacting and noninteracting glycopeptides. Previous studies have shown that most of the complex-type Asn-linked oligosaccharides in BW5147 cells are bi-, tri-, and tetraantennary species (17, 20, 21). Many of the tri- and tetraantennary species (designated glycopeptides B2 and A2, respectively) contain the repeating disaccharide (3Gal@l,4GlcNAc@) or poly-n-acetyllactosamine sequence (17, 21). Many of the penultimate galactose residues in these glycopeptides are sialylated either a2,3 or a2,6; however, the latter type of linkage is the predominant type (17). Some of the tri- and tetraantennary complex-type glycopeptides (designated B1 and A1, respectively) lack the poly-n-acetyllactosamine sequence and contain small amounts of either galactose or sialic acid (20, 21). Fewof the biantennary glycopeptides (designated A) contain the poly-n-acetyllactosamine sequence, and most of these contain sialic acid in a2,6-linkage (20, 21). To investigate the interactions of glycopeptides from BW5147 cells with MAL-Sepharose, we prepared radiolabeled glycopeptides from these cells. The cells were metabolically radiolabeled with either 13H]galactose or [3H]mannose precursors, and the radiolabeled glycopeptides A1,A2, B1, B2, and A were isolated by chromatography on immobilized lectins, as described under Experimental Procedures. The elution profile of [3H]mannose-labeled A1, B1, A2, and B2 on MAL-Sepharose is shown in Fig. 3. A significant fraction of glycopeptides A2 and B2 were retarded in their elution from MAL-Sepharose; in contrast, only a small portion of glycopeptide B1 and no significant amount of glycopeptide A1 was retarded. We recovered the A2 glycopeptides that were not retarded in their elution from MAL-Sepharose (designated A2Ma) and those that were retarded in their elution from the column (designated A2Mb) (Fig. 3c). The A2Mb glycopeptides were reapplied to the MAL-Sepharose and were again clearly retarded in their elution from the column (Fig. 4a). When these glycopeptides were desialylated by treatment with either mild acid (Fig. 46) or Arthrobacter neuraminidase (Fig. 4c), the 50 A2Ma ALMC b n \ F RAC T ON FG. 3. Chromatography of glycopeptides on MAL-Sepharose. Glycopeptides A1, B1, A2 and B2 (ad, respectively) from [3H]mannose-labeled BW5147 cells were chromatographed on MAL- Sepharose as described under Experimental Procedures. Retarded (ZA2Mb) and unbound (A2Ma) glycopeptides were pooled as indicated in panel c. One-ml fractions were collected, and the recovery of glycopeptides was greater than 90%. The elution volume of unbound and noninteractive material was approximately 4.5 ml and was contained largely in fraction number 5.

6 nteractions of Siulylated Glycoconjugates with mmobilized MAL 4581 E o c d \.e FRACT ON FG. 4. Chromatography of glycopeptides A2Mb on MAL- Sepharose before and after desialylation. The [3H]mannoselabeled A2Mb glycopeptides were analyzed by chromatography on MAL-Sepharose before (a) and after treatment with mild acid treatment (b), Arthrobacter neuraminidase treatment (c), and heat-inactivated neuraminidase (d). The elution of unbound and noninteractive material was approximately 4.5 ml and was contained largely in fraction number 5. detectable interaction of the glycopeptides with the lectin was abolished. Treatment of the A2Mb glycopeptides with either a-fucosidase, or P-N-acetylhexosaminidase did not alter the interaction of the glycopeptides with the immobilized lectin (data not shown). These results demonstrate that sialic acid determinants on certain subsets of [3H]mannose-labeled glycopeptides from the BW5147 cells are required for high affinity interactions with immobilized MAL. Structural Analyses of the A2Ma and A2Mb Glycopeptides-We have recently shown that many of the po1y-nacetyllactosamine sequences in the tri- and tetraantennary complex chains from BW5147 cells contain terminal sialic acid residues in either a2,3- or a2,6-linkage to galactose (17). These poly-n-acetyllactosamine chains are released from the glycopeptides by treatment with E. freundii endo-p-galactosidase giving rise to five species of oligosaccharides, which have the following structures: Siaa2,3Gal~l,4GlcNAc~l,3Gal; sidase elute in the region of peak 1 and the neutral di-, tri-, and tetrasaccharide species released by the enzyme elute in Siaa2,6Gal~l,4GlcNAc~l,3Gal; Galal,3Gal~l,4GlcNAc~l, the region of peak Gal; Gal/31,4GlcNAc(31,3Gal; and GlcNAcpl,3Gal (17, 21). Since the two sialylated oligosaccharides released by end0-pgalactosidase are easily separable by descending paper chromatography (17), we used this technique to assess the sialylation pattern of glycopeptides bound by MAL-Sepharose. Peak 11, derived from endo-p-galactosidase treatment of the A2Ma glycopeptides, was analyzed by descending paper chromatography (Fig. 5). Greater than 90% of the radioactivity co-migrated with the tetrasaccharide Siaa2,6Ga1/31,4Glc- NAcP1,3Gal (Fig. 5). Peak 1 derived from endo-p-galactosid- We prepared and isolated metabolically labeled [3H]galac- ase-treated A2Mb glycopeptides was also analyzed by detose-labeled A2Ma and A2Mb glycopeptides from BW5147 scending paper chromatography (Fig. 6). Two major and cells and treated these glycopeptides with E. freundii endo-@- galactosidase. The treated A2Ma and A2Mb samples were applied to a column of Bio-Gel P-10, as shown in Figs. 5 and 6, respectively. Prior to treatment, both A2Ma and A2Mb glycopeptides eluted near the void volume of the column (data not shown). However, treatment of both glycopeptides with the enzyme gave rise to three major peaks of radioactivity upon column chromatography on Bio-Gel P-10. The peaks were designated, 11, and 111 (Figs. 5 and 6) and the fractions were pooled, dried, and further analyzed. Peak material from treatment of both A2Ma and A2Mb represents residual glycopeptide and was not analyzed further. Peaks 1 and 111, which contained the [3H]galactose-labeled oligosaccharides released by the enzyme treatment of both 200- O O CM FG. 5. Chromatography of [3H]galactose-labeled A2Ma glycopeptides before and after treatment with E. freundii endo-&galactosidase. The [3H]galactose-labeled A2Ma glycopeptides were treated for 48 h with endo-0-galactosidase. The treated glycopeptides were applied to a column of Bio-Gel P-10 (uppergraph), which resulted in three peaks of radiolabeled material designated peaks Z, ZZ, and ZZ. Fractions in these peaks were pooled together as shown. The oligosaccharides in peaks ZZ and ZZZ were analyzed by preparative descending paper chromatography (lower graphs), as indicated by the arrows between graphs. The migration of standards are indicated by the arrows; 1, Siaa2,6Gal~l,4GlcNAc~l,3Gal; Siaa2,3Gal01,4GlcNAc01,3Gal; 3, Galal,3Gal~1,4GlcNAc~l,3Gal; 4, Gal~1,4GlcNAc01,3Gal; 5, GlcNAc,!301,3Gal. Fractions containing the oligosaccharides in peaks a, a, b, and c were pooled as indicated. The V, and V, of the Bio-Gel P-10 column are shown and were determined by the elution position of blue dextran and mannose, respectively. the A2Ma and A2Mb glycopeptides, represented approximately 55% of the total initial [3H]galactose label in the glycopeptides. Previous studies (17) have shown that the monosialylated oligosaccharides released by endo-@-galacto- nearly equal peaks of radioactivity were obtained; one which co-migrated with Siaa2,6Gal~l,4GlcNAc~l,3Gal, and one which co-migrated with Siaa-2,3Gal~l,4GlcNAc~l,3Gal (Fig. 6). These results indicate that the poly-n-acetyllactosamine sequences on the A2Ma glycopeptides, not bound by MAL- Sepharose, contain sialic acid predominantly in a2,6-linkage to galactose, whereas many of these chains on the A2Mb glycopeptides which interact with the MAL-Sepharose contain sialic acid-linked a2,3 to galactose. The neutral oligosaccharides contained in peak 111, derived from endo-p-galactosidase treatment of A2Ma and A2Mb glycopeptides, were analyzed by preparative descending paper chromatography (Figs. 5 and 6). Three peaks of material were

7 4582 nteractions of Sialylated Glycoconjugates with mmobilized MAL n 0 vo # r V v FG. 6. Chromatography of [SH]galactose-labeled A2Mb glycopeptides before and after treatment with E. freundii endo-@-galactosidase. The [3H]galactose-labeled APMb glycopeptides were treated for 48 h with endo-@-galactosidase. The treated glycopeptides were applied to a column of Bio-Gel P-10 (uppergraph), resulting in three peaks of radiolabeled material designated peaks, Z, and ZZ. Fractions in these peaks were pooled together as shown. The oligosaccharides in peaks 11 and Z were analyzed by preparative descending paper chromatography (lower graphs), as indicated by the arrows between the graphs. The standards 1-5 used for descending paper chromatography are the same as shown in Fig. 5. Fractions containing the oligosaccharides in peaks a, b, a, b, and c were pooled as indicated. The V, and V. of the Bio-Gel P-10 column are shown and were determined by the elution position of blue dextran and mannose, respectively. that peak a is the tetrasaccharide Gala1,3Gal/31,4GlcNAc- /31,3Gal; peak b is the trisaccharide Gal/31,4Glc- NAc/31,3Gal; and peak c is the disaccharide GlcNAc/31,3Gal. The sialylated oligosaccharides in peak 1 (Figs. 5 and 6) were desialylated by mild acid treatment and analyzed directly by HPLC. Only one peak of radioactive material wasrecovered in each sample, and that material eluted with the same retention time as the standard Ga1/31,4GlcNAc/31,3Gal (data not shown). n summary, the results demonstrate that peak a is the tetrasaccharide Siaa2,6Gal/31,4GlcNAcGl, 3Gal and that peak b is the tetrasaccharide Siaa2,3Gal/31,4GlcNAc/31,3- Gal. The structures of these oligosaccharides were identical to those previously found in glycopeptides from BW5147 cells (17, 21). n addition, these results indicate that the A2Ma and A2Mb glycopeptides contain similar structures except that they differ in the type of sialylation. To further investigate the degree and heterogeneity of sialylation of the A2Mbglycopeptides, the [3H]mannoselabeled material was analyzed by ion exchange column chromatography on QAE-Sephadex (Fig. 8). Previous studies have shown that glycopeptides containing one negative charge are bound by QAE-Sephadex and eluted with 20 mm NaCl, whereas glycopeptides containing two and three charges are eluted with 70 and 140 mm NaC1, respectively (30). When the A2Mb glycopeptides were analyzed on QAE-Sephadex, a majority of the [3H]mannose-labeled material was eluted with 70 and 140 mm NaCl (Fig. 8a). Treatment of the A2Mb glycopeptides with either mild acid or neuraminidase, as described under Experimental Procedures, reduced the net charge of the glycopeptides. Most of the desialylated glycopeptides were bound by QAE-Sephadex, but were eluted with 20 mm NaCl (Fig. 8, b and c). These results suggest that the A2Mb glycopeptides are all sialylated, and some glycopeptides contain 1 and some contain 2 sialic acid residues. The residual charges on the desialylated glycopeptides are due to peptide moieties. To determine the branching pattern of the complex-type Asn-linked oligosaccharides interacting with high affinity with MAL-Sepharose, we methylated the [3H]mannose-labeled B2 and A2 glycopeptides. Methylation of B2 glyco- FRACTON FG. 7. HPLC of [SH]galactose-labeled oligosaccharides released from A2Mb glycopeptides by endo-@-galactosidase. Portions of the oligosaccharides recovered in peaks a, b, and c of the paper chromatogram shown in Fig. 6 were mixed and analyzed by HPLC as described under Experimental Procedures. The elution positions of the oligosaccharide standards is indicated by their structures and correspond from left to right to G-g (GlcNAc@l,gGal), g-gg (Gal~1,4GlcNAc@1,3Gal),andg-g-G-g (Galal,3Gal@l,4GlcNAc@l,3- Gal). recovered; the slower migrating peak, a, co-migrated with Gala1,3Gal/3l,4GlcNAc/31,3Gal; the next faster migrating peak, b, co-migrated with Gal/31,4GlcNAc/31,3Gal; and the fastest migrating peak, c, co-migrated with GlcNAc/31,3Gal. To further examine this material, portions of the samples recoveredfrom the three peaks by preparative descending paper chromatography were analyzed either individually by HPLC (data not shown) or as a prepared mixture (Fig. 7). The three peaks of material obtained had elution times identical to the indicated standards. These results demonstrate 1ooc. n FRACTON FG. 8. QAE-Sephadex column chromatography of [ H mannose-labeled A2Mb glycopeptides before and after desialylation. a, a portion of the [3H]mannose-labeled A2Mb glycopeptides was applied to a column of QAE-Sephadex and the column eluted with increasing concentrations of NaC, as described under Experimental Procedures; b, chromatography of the glycopeptides after treatment with mild acid; c, chromatography of the glycopeptides after treatment with neuraminidase; d, chromatography of glycopeptides after treatment with inactivated neuraminidase.

8 nteractions of Sialylated Glycoconjugates with mmobilized MAL 4583 peptides gave rise to the following methylated mannose and fucose residues: 3,4-di-O-methylmannitol, 2,4-di-O-methylmannitol and 3,4,6-tri-O-methylmannitol, and 2,3,4-tri-Omethylfucitol in the ratio of l.ol.o:l.o:l.o, respectively (Fig. 9a). These results indicate that this glycopeptide consists exclusively of triantennary Asn-linked chains with one a- linked mannose substituted at C-2 and the other substituted at both C-2 and C-6. The residue of 2,4-di-O-methylmannitol recovered upon methylation is probably derived from linked residue of mannose in the core of all Asn-linked sugar chains which is substituted at positions C-3 and C-6 by the a-linked mannose residues (10). Triantennary oligosaccharides with this branching pattern and a residue of fucose attached a1,6 to the GlcNAc linked to Asn are known to bind with high affinity to immobilized pea lectin (2, 37). Methylation of the A2 glycopeptides gave rise to the following four methylated mannose residues: 3,4-di-O-methylmannitol, 3,6-di-O-methylmannitol, 2,4-di-O-rnethylmannitol and 3,4,6-tri-O-methylrnannitol, and 2,3,4-tri-O-methylfucitol, in the ratio of 0.8:0.81.O0.6:1.0, respectively (Fig. 9b). These results indicate that A2 glycopeptide consists of a mixture of tri- and tetraantennary Asn-linked chains. Methylation analysis of the [3H]mannose-labeled A2Ma and A2Mb glycopeptides gave similar results to that shown for the total A2 glycopeptides (data not shown). These results indicate that the glycopeptides A2Ma and A2Mb do not differ significantly in the type of branching patterns. Enzymatic Resialylntion of the Desiulylnted Derivatives of Glycopeptide A2Mb and A2Ma As indicated above, the major structural difference between the A2Mb and A2Ma glycopeptides is the linkage of sialic acid residues to galactose. The A2Mb glycopeptides contain sialic acid-linked a2,3 to galactose, whereas the A2Ma glycopeptides contain sialic acid-linked a2,6 to galactose. These results indicate that the linkage of sialic acid a2,3 to galactose is the primary structural determinant required for high affinity interactions with immobilized MAL. To further test this possibility, we desialylated both the [3H]galactose-labeled A2Mb and A2Ma glycopeptides by treatment with mild acid; the desialylated glycopeptides were then enzymatically resialylated by incubation with either purified CMPNeuAc:Gal@(l,3/4)GlcNAc a2,3-sialyltransferase from rat liver (35, 36) or purified CMPNeuAc: Gal@l,4GlcNAc a2,6-~ialyltransferase from rat liver (35, 36), as described under Experimental Procedures. The efficiency of resialylation of the asialo derivatives of the A2Ma and A2Mb glycopeptides was examined by column chromatography of the glycopeptides on QAE-Sephadex (Figs n R FG. 10. QAE-Sephadex column chromatography of [ Hlgalactose-labeled A2Ma glycopeptides and enzymatically resialylated A2Ma glycopeptides. The [3H]galactose-labeled A2Ma glycopeptides were desialylated by mild acid treatment and enzymat- ically resialylated by purified a2,3-~ialyltransferase and a2,6-sialyltransferase, as described under Experimental Procedures. The glycopeptides were applied to a column of QAE-Sephadex as described in the legend to Fig. 8. a, native A2Ma glycopeptides; b, desialylated AQMa glycopeptides; c, desialylated A2Ma glycopeptides resialylated by the a2,3-~ialyltransferase; d, desialylated A2Ma glycopeptides resialylated by the a2,6-sialyltransferase. b FG. 9. Separation of radiolabeled methylated mannose and fucose residues by reverse-phase HPLC. [3H]Mannose-labeled B2 glycopeptides (a) and A2 glycopeptides (b) from BW5147 cells were subjected to methylation, hydrolysis, and reduction. The methylated mannose and fucose derivatives were separated by HPLC as described under Experimental Procedures. Peaks 1-5 correspond to 3,4-di-O-methylmannitol, 3,6-di-O-methylmannitol, 2,4-di-0- methylmannitol, 3,4,6-tri-O-methylmannitol, and 2,3,4-tri-O-rnethylfucitol, respectively FRACTON FG. 11. QAE-Sephadex column chromatography of [ Hlgalactose-labeled A2Mb glycopeptides and enzymatically resialylated A2Mb glycopeptides. The [3H]galactose-labeled A2Mb glycopeptides were desialylated by mild acid treatment and enzymatically resialylated by purified a2,3-sialyltransferase and a2,6-sialyltransferase, as described under Experimental Procedures. The glycopeptides were applied to a column of QAE-Sephadex as described in the legend to Fig. 8. a, native A2Mb glycopeptides; b, desialylated A2Mb glycopeptides; c, desialylated AQMb glycopeptides resialylated by the a2,3-sialyltransferase; d, desialylated A2Mb glycopeptides resialylated by the a2,6-sialyltransferase.

9 4584 nteractions of Sialylated Glycoconjugates with mmobilized MAL observed that leukoagglutination by MAL was inhibited by low concentrations of 2,3-sialyllactose, while it was not inhibited by much higher concentrations of 2,6-sialyllactose (Table ). We also found that high concentrations of N-acetylneuraminic acid were ineffective in inhibiting leukoagglutination, which is consistent with the finding that the mitogenic activ ity ofmal is not inhibited by high concentrations of N- acetylneuraminic acid (15). These results indicated that MAL FRACTON (0.5ml) interacts with high affinity with glycoconjugates containing FG. 12. Chromatography of enzymatically resialylated [3H] NeuAc-linked a2,3 to galactose. galactose-labeled glycopeptides A2Ma and A2Mb on MAL- To investigate in more detail the sialylated sequences pro- Sepharose. Glycopeptide fractions A2Ma and A2Mb were treated with mild acid and enzymatically resialylated by either the purified moting high affinity interactions with MAL, we chose to a2,3-~ialyltransferase or the purified a2,6-sialyltransferase. The examine the interaction of glycopeptides of differing strucresialylated glycopeptides were then applied to a column of MAL- tures with immobilized MAL. Our results demonstrate that Sepharose, as described under Experimental Procedures. a, desialylated A2Ma glycopeptides resialylated with the a2,3-sialyltransferase; b, desialylated A2Mb glycopeptides resialylated with the a2,3- sialyltransferase; c,desialylateda2maglycopeptidesresialylated with the a2,6-sialyltransferase; and, desialylated A2Mb glycopep- tides resialylated with the a2,6-sialyltransferase. The elution volume of unbound and noninteractive material was approximately 4.5 ml and was contained in fraction numbers 9 and and 11). Enzymatic resialylation of the asialo derivatives of the glycopeptides by either enzyme gave rise to glycopeptides with elution patterns on QAE-Sephadex similar to that of the native glycopeptides (Figs. 10 and 11). These results do not indicate that resialylation was complete but rather demonstrate that a majority of the glycopeptides were resialylated by the enzymes. We then examined the interactions of the resialylated glycopeptides with immobilized MAL (Fig. 12). Glycopeptides that were resialylated with a2,6-sialyltransferase were not noticeably retarded in their elution from the column (Fig. 12, c and d). However, a significant percentage of the glycopeptides that were resialylated with the a2,3-~ialyltransferase were retarded in their elution from the column (Fig. 12, a and b). The fact that some of the A2Ma and A2Mb glycopeptides that were resialylated with the a2,3-sialyltransferase were not retarded in their elution from the MAL-Sepharose is most likely due to incomplete resialylation of the glycopeptides by the enzyme. The most important observation is that the A2Ma glycopeptides, which originally contained the terminal sequences Siaa2,6Gal-R, could interact with high affinity with MAL- Sepharose when the terminal sequence was enzymatically modified to Siaa2,BGal-R (Fig. 12a). These results provide additional evidence that the presence of the terminal sequence Siaa2,3Gal@1,4GlcNAc-R on complex-type tri- and tetraantennary Asn-linked oligosaccharides is required for high affinity interactions of glycopeptides with immobilized MAL. DSCUSSON Our earlier observation that the leukoagglutinin (MAL) from M. arnurensis seeds is a very weak hemagglutinin and a potent leukoagglutinin toward the BW5147 mouse lymphoma cells (14) prompted us to investigate the carbohydrate binding specificity of the lectin. Kawaguchi et al. (15) originally observed that MAL is a potent mitogen for peripheral blood lymphocytes and a poor hemagglutinin. n the same study, however, they did not obtain detailed information about the carbohydrate binding specificity of the lectin using the available reagents to inhibit mitogenicity. We initially chose to investigate the carbohydrate binding specificity of MAL using a newly developed leukoagglutination assay (14) to assess the ability of several purified oligosaccharides to inhibit leukoagglutination by the lectin. We complex-type tri- and tetraantennary Asn-linked oligosaccharides that contain the terminal sequence Siaa2,3Galp1,4- GlcNAc-R interact with high affinity with immobilized MAL and can be efficiently separated from glycopeptides that lack this sequence (Figs. 1-3). t is interesting to note that, although MAL is a potent leukoagglutinin for BW5147 cells and interacts with high affinity with sialylated glycoconjugates, the lectin is a weak hemagglutinin. This is likely due to differences in the sialylation of surface glycoconjugates between BW5147 cells and human erythrocytes. Much of the erythrocyte sialic acid is contained in glycophorin, which contains sialic acid in both Asn-linked and Ser/Thr-linked oligosaccharides (38,39). The bisected biantennary complex-type Asn-linked oligosaccha- rides of glycophorin contain the terminal sequence NeuAca2,6Gal~l,4GlcNAc~l,2Man-R (39). Our results indicate that this type of sialylated sequence is not bound with high affinity by MAL. However, the sialylated 0-linked sugar chains of glycophorin consist of NeuAca2,3Gal~1,3Gal- NAcal-R and NeuAca2,3Galpl,3 (NeuAca2,6)GalNAcal-R (38). Thus, it appears that MAL binds with low affinity to the sialylated 0-linked chains of glycophorin, even though these chains contain sialic acid-linked a2,3 to galactose. These data suggest that the binding site of the lectin is complex and may recognize, in addition to a2,3-linked sialic acid, aspects of the underling oligosaccharide sequence. Our studies indicate that sialylated poly-n-acetyllactosamine sequences are not required for high affinity interactions of glycopeptides with immobilized MAL. Many of the complex-type chains from BW5147 cells bound by MAL-Sepharose contain sialylated poly-n-acetyllactosamine sequences, and these sequences are not present in fetuin-derived glyco- peptide. n addition, our results indicate that the high affinity interactions are not dependent of the branching pattern of the mannose residues in the complex-type chains. The complex-type tri- and tetraantennary Asn-linked oligosaccharides from the BW5147 cells interacting with high affinity with immobilized MAL contain a branching structure on the mannose residues different than that for the complex-type chain from fetuin, which also interacts with MAL. Thus, it is likely that a major structural determinant promoting high affinity interactions of glycopeptides with MAL-Sepharose is the presence of the sequence Siaa2,3Galpl,4GlcNAc-R. We have observed that intact fetuin glycopeptide is bound by MAL-Sepharose, and elution requires the application of buffer having a low ph (ph 3.0), whereas desialylated fetuin

10 nteractions of Sialylated Glycoconjugates with mmobilized MAL 4585 is not bound by the immobilized lectin (data not shown). t should be recalled that the glycopeptides derived from fetuin are retarded and fractionated by MAL-Sepharose (Fig. 2b). t is possible that fractionation of the fetuin glycopeptides is due to heterogeneity and degree of sialylation. These results strongly suggest that the density of sialylated residues might enhance the affinity of glycoconjugates for MAL. n addition, higher concentrations of immobilized MAL on the column would probably promote stronger binding of appropriately sialylated glycopeptides. Lectins which bind to sialic acid on glycoconjugates have been isolated from plants and many other organisms. These lectins include wheat-germ agglutinin (40,41); elderberry bark lectin (12, 13); sialic acid binding proteins in Streptococcus sanguis (42) and Streptococcus mitis (43); the agglutinin in red algae Solierin choradalis (44); the sialic acid binding lectins in Mmtigoproctus giganteus (45), Limulus polyphemus (46), Aphonopelma spiders (47), the horseshoe crab Carcinoscorpius rotunda cauda (48-50), the slug LimuxfZuuus (51,52), and the snail Cepaea hortensis (53). A lectin which binds specifically to 0-acetylated sialic acids has been purified from the marine crab Cancer antenrzurius (54). n addition, the hemagglutinin of influenza C virus binds with high affinity to 9-0-acetylneuraminic acid (55). Of direct importance to our study, however, are the recent reports demonstrating that the elderberry bark lectin (Sambucus nigra L.) binds glycoconjugates containing the terminal sequence Siaa2,6Gal@1,4GlcNAc-R, while those isomers containing sialic acid-linked a2,3 are not bound (12, 13). t is important to note that this lectin is a hemagglutinin, whereas MAL is not a hemagglutinin. We have now shown that MAL binds with high affinity to sialylated oligosaccharides containing the sequence SiaaZ,BGa1@1,4GlcNAc-R. Thus, the combined use of the immobilized forms of both the elderberry bark lectin and MAL in affinity chromatographic techniques should now permit more complete fractionation of sialylated glycoconjugates on the basis of sialic acid linkages, and thereby greatly facilitate the isolation and structural analysis of animal cell glycoconjugates. Acknowledgments-We thank Dr. James Paulson and his colleaguesforkindly resialylating glycopeptides with purified sialyltransferases, and we thank Dr. David Smith for supplying purified oligosaccharides for this study. REFERENCES 1. Lis, H., and Sharon, N. (1986) Annu. Reu. Biochem. 56, Merkle, R. K., and Cummings, R. D. (1987) Methods Enzymol. 138, Barondes, S. H. (1981) Annu. Reu. Biochem. 60, Blake, D. A., and Goldstein,. J. (1982) Methods Enzymol. 83, Finne, J., and Krusius, T. (1982) Methods Enzymol. 83, Lis, H., and Sharon, N. (1984) in Biology of Carbohydrates (Ginsburg, V., and Robbins, P. W., eds) Vol. 2, pp. 1-85, John Wiley and Sons, New York 7. Sadler, J. E. (1984) in Biology of Carbohydrates (Ginsburg, V., and Robbins, P. W., eds) Vol. 2, pp , John Wiley and Sons, New York 8. Kobata, A. (1984) in Biology of Carbohydrates (Ginsburg, V., and Robbins, P. W., eds) Vol. 2, pp , John Wiley and Sons, New York 9. Snider, M. D. (1984) in Biology of Carbohydrates (Ginsburg, V., and Robbins, P. W., eds) Vol. 2, pp , John Wiley and Sons, New York 10. Kornfeld, R., and Kornfeld, S. (1985) Annu. Reu. Biochem. 64, Schauer, R. (1985) Trends Bwchem. Sei. 10, Shibuya, N., Goldstein,. J., Broekaert, W. F., Nsimba-Lubaki, M., Peeters, B., and Peumans, W. J. (1987) J. Biol. Chem. 262, Shibuya, N., Goldstein,. J., Broekaert, W. F., Nsimba-Lubaki, M., Peeters, B., and Peumans, W. J. (1987) Arch. Biochem. Biophys. 254, Wang, W.-C., and Cummings, R. D. (1987) Anal. Biochem Kawaguchi, T., Matsumoto,., and Osawa, T. (1974) J. Biol. Chem. 249, Li, Y.-T., and Li, S.-C. (1972) Methods Enzymol. 28, Merkle, R. K., and Cummings, R. D. (1987) J. Biol. Chem. 262, Laemmli, U. K. (1970) Nature 227, Lowry, 0. H., Rosebrough, N. J., Farr, A. L., and Randall, R. J. (1951) J. Biol. Chem. 193, Cummings, R. D., and Kornfeld, S. (1982) J. Biol. Chem. 257, Cummings, R. D., and Kornfeld, S. (1984) J. Biol. Chem. 259, Spiro, R. G., and Bhoyroo, V. D. (1974) J. Biol. Chem. 249, Yamamoto, K., Tsuji, T., rimura, T., and Osawa, T. (1981) Biochem. J. 195, Cummings, R. D., and Kornfeld, S. (1982) J. Biol. Chem. 257, Dubois, M., Gilles, K. A., Hamilton, J. K., Rebers, P. A., and Smith, F. (1956) Anal. Chem. 28, Baenziger, J. U., and Fiete, D. (1979) J. Bid. Chem. 254, Nilsson, B., Nordbn, N. E., and Svensson, S. (1979) J. Biol. Chem. 264, Takasaki, S., and Kobata, A. (1986) Biochemistry 26, Finne, J., and Krusius, T. (1982) Methods Enzymol. 83, Varki, A,, and Kornfeld, S. (1983) J. Biol. Chem. 258, Hakomori, S. (1964) J. Biochem. (Tokyo) 65, Li, E., Tabas,., and Kornfeld, S. (1978) J. Biol. Chem. 253, Stoffyn, P., Stoffyn, A,, and Hauser, G. (1971) J. Lipid Res. 12, Szilagyi, P. J., Arango, J., and Pierce, M. (1985) Anal. Biochem. 148, Weinstein, J., de Souza-e-Silva, U., and Paulson, J. C. (1982) J. Biol. Chem. 257, Weinstein, J., de Souza-e-Silva, U., and Paulson, J. C. (1982) J. Biol. Chem. 267, Kornfeld, K., Reitman, M. L., and Kornfeld, R. (1981) J. Biol. Chem. 256, Thomas, D. B., and Winzler, R. J. (1969) J. Biol. Chem. 244, Yosbima, H., Furthmayr, H., and Kobata, A. (1980) J. Bid. Chem. 255, Kronin, K. A., and Carver, J. P. (1982) Biochemistry 21, Furukawa, K., Minor, J. E., Hegarty, J. D., and Bhavanandan, V. P. (1986) J. Biol. Chem. 261, Murray. P. A., Levine, M. J.. Tabak. L.A.. and Reddv. M. S. (1982) Biochem. Biophys. Res. Commun. 106, Murray, P. A., Levine, M. J., Reddy, M. S., Tabak, L. A., and Bergey, E. J. (1986) nfect. mmunol. 53, Rogers, D. J., and Topliss, J. A. (1983) Bot. Mar. 26, Vasta, G. R., and Cohen, E. (1984) J. nuert. Pathol. 43, Muresan, V., wanij, V., Smith, Z. D. J., and Jamieson, J. D. (1982) J. Histoehem. Cvtochem Vasta, G. R., agd Cohen, E.'(1984) Deu. Comp. mmunol. 8, Thambi Dorai, D., Bachhawat, B. K., Bisbayee, S., Kannan, K., and Rao, -, ~~, D. R. (1980) Arch. Biochem. Biophys. 209, Bisbayee, S., and Thambi Dorai, D. (1980) Biochim. Biophys. Acto 623, Mohan, S., Thambi Dorai, D., Srimal, S., and Bacbhawat, B.K. (1982) Biochem. J. 203, Miller, R. L., Collawn, J. F., and Fish, W. W. (1982) J. Biol. Chem. 257, Miller, R. L. (1982) J. nuert. Pathol. 39, Holm, S. E., Bergholm, A.-M., Wagner, B., and Wagner, M. (1985) J. Med. Microbiol Ravindranath, M. H., Higa, H. H., Cooper, E. L., and Paulson, J. C. (1986) J. Biol. Chem. 260, Rogers, G. N., Herder, G., Paulson, J. C., and Klenk, H-D. (1986) J. Bid. Chem. 261, Townsend, R. R., Hardy, M. R., Wong, T. C., and Lee, Y. C. (1986) Biochemistry 25,