Structural and conformational features that affect the interaction of polyactosaminoglycans with immobilized wheat germ agglutinin

Biochimica et Biophysica Acta 883 (1986) 253-264 Elsevier

253

BBA 22241

S t r u c t u r a l and c o n f o r m a t i o n a l f e a t u r e s that a f f e c t the i n t e r a c t i o n of p o l y l a c t o s a m i n o g l y c a n s with i m m o b i l i z e d w h e a t g e r m agglutinin R a y m o n d J. I v a t t *, J o h n W. R e e d e r a n d G a r y F. C l a r k Department of Tumor Biology, The University of Texas, M.D. Anderson Hospital and Tumor Institute at Houston, Houston, TX 77030 (U.S.A.) (Received January 28th, 1986)

Key words: Wheat germ agglutinin; Polylactosaminoglycan; Erythroglycan; Glycan-lectin interaction; Lectin; Glycan

We examined the interaction between immobilized wheat germ agglutinin and the large, polylactosaminecontaining glycans from human erythrocytes and human K-562 erythroleukemic cells. Three classes of interaction were identified. One class of glycan was merely retarded during chromatography. The other two classes were retained and could be distinguished by their ease of displacement with N-acetylglucosamine (GIcNAc); one was a moderate-affinity fraction displaced by 0.1 M GicNAc and the other was a high-affinity fraction subsequently displaced by 1.0 M GlcNAc. A relatively small fraction of the K-562 polylactosamines were in the high-affinity class. We explored the role that fucose and sialic acid substitutions play in the strength of the lectin-glycan interaction. Although sialic acid is recognized by wheat germ agglutinin, sialylation was not required for the high-affinity interaction, and the presence of sialic acids actually prevented some glycans from binding with high affinity. In contrast, fucose is not part of the binding determinant, yet the removal of fucose resulted in decreased affinity. The possibility that some of these changes in affinity were the result of conformational changes was explored using matrices that had wheat germ agglutinin immobilized at different densities. At low wheat germ agglutinin densities, adult and fetal erythroglycans and K-562 glycophorin-like glycans were not retained by the matrix. As the density increased, the proportion of glycans that were retarded, and ultimately retained, increased. While these increases in the proportions retained occurred in parallel for the three different glycans, the apparent affinities of the glycan-lectin interactions differed. The glycophorin-like glycans were always readily displaced by 0.1 M GIcNAc, even at higher wheat germ agglutinin densities. In contrast, as the wheat germ agglutinin density increased, the proportion of erythroglycans that could be displaced by 0.1 M GIcNAc decreased; at 10 m g / m l immobilized wheat germ agglutinin, greater than 80% of the erythroglycans exhibited this tighter interaction. We suggest that this higher affinity interaction is the result of the large glycans spanning adjacent wheat germ agglutinin molecules, and is determined by the proximity of these molecules and the conformation of the glycans. Introduction Polylactosaminoglycans are relatively large, repeating copolymers of the sugars galactose and * To whom correspondence should be addressed. This work was supported by Grant 1-972 from the March of Dimes Birth Defects Foundation and Grant CA42650 from the National Cancer Institute.

N-acetylglucosamine (GlcNAc) reported to be present on embryonic, fetal, and neoplastic cell types [1-3]. In the normal adult they are expressed on the cells of the reticuloendothelial system. Their restricted distribution and developmental regulation suggests an involvement in transient cellular interactions [4]. These polymeric carbohydrates may be attached to asparagine-linked mannose

0304-4165/86/$03.50 © 1986 Elsevier Science Publishers B.V. (Biomedical Division)

254

cores [5], serine-linked oligosaccharides or ceramide [6]. In addition, these glycans may be further decorated with fucose, sialic acid, or N-acetylgalactosamine [5]. Two major classes of polylactosaminoglycans have been identified: linear polymers possessing Gal-/31-4-GIcNAc units linked exclusively by /31-3 linkages and branched polymers having additional ill-6 linkages [7]. These linear and branched glycans are directly responsible for i and I blood group specificities [8] and have been demonstrated to be the major carriers for ABO blood groups on human erythrocytes [9,10]. The information now known about polylactosaminoglycan structure has been obtained by isolating and characterizing carbohydrates derived from glycoproteins that express substantial amounts of these glycans, such as erythrocyte band 3 glycoprotein [5,11]. While such investigations have provided a general understanding of polylactosamine structure, they have described average structures present in heterogeneous populations of glycans. Despite the development of antibodies to polylactosaminoglycans, antibodycoupled affinity columns have not yet proved useful for the isolation and resolution of these carbohydrates. It has recently been reported that polylactosaminoglycans can bind with very high affinity to immobilized wheat germ agglutinin [12,13]. High affinity was for the polylactosaminoglycans from human erythrocytes (erythroglycans), defined as retention by the immobilized lectin in the presence of 0.1 M GIcNAc buffer and subsequent elution with either 3 mM tri-N-acetylchitotriose or 1.0 M GIcNAc buffer. These erythroglycans are a mixture of branched and linear polylactosamines. We have extended these studies by examining the interaction of the linear polylactosamines from K562 human erythroleukemic cells with immobilized wheat germ agglutinin. We have prepared different affinity fractions based upon their relative ease of displacement from the lectin by Nacetylglucosamine. The effects upon the lectinglycan interaction of various chemical and enzymatic treatments and of having wheat germ agglutinin immobilized at different densities have been examined. These experiments indicate that the binding of linear polylactosaminoglycans to

immobilized wheat germ agglutinin is very complex, with the higher-affinity interactions determined by the proximity of the wheat germ agglutinin molecules and the conformation of the glycans. Materials and Methods

Materials. Endo-fl-galactosidase from Escherichia freundii and galactose oxidase/NaB(3H)4labeled erythrocyte band 3 glycopeptide standards were the generous gifts of Drs. Michiko and Minoru Fukuda. [6-3H]Glucosamine (21 C i / mmol) and [1-3H]galactose (25 Ci/mmol) were obtained from ICN, Irvine, CA. Bio-Rad Laboratories, Richmond, CA, was the supplier for Bio-Gel P4, Bio-Gel P10, and Affigel 10. Wheat germ agglutinin-Sepharose 6MB and QAE-Sephadex were obtained from Pharmacia Fine Chemicals, Piscataway, NJ. Wheat germ agglutinin was obtained from E-Y Laboratories, San Mateo, CA. Antisera against the iI blood group was generously provided by Dr. John Moulds of Gamma Biologicals, Houston, TX. Neuraminidase (Clostridium perfringens) was obtained from Sigma Chemical Co., St. Louis, MO. Cell culture. K-562 cells were kindly donated by Dr. C. Reading of our department. The cells were maintained in Dulbecco's modified Eagle's medium with 10% fetal calf serum (Grand Island Biological Company, Santa Clara, CA). Metabolic labeling and extraction. K-562 cells were grown in suspension culture at 37°C in an atmosphere of 5% CO 2 and air. The ceils were maintained in exponential growth phase at a cell density of (2-5)-105 cells/ml and then were harvested and centrifuged for 5 min at 1000 × g. After being washed with 10 ml phosphate-buffered saline, 107 cells were incubated in 30 ml Dulbecco's .modified Eagle's medium containing either [3H] glucosamine or [3H]galactose (each at 10 #Ci/ml) and 10% fetal calf serum. The K-562 cells were maintained at 37°C for 15 h and were centrifuged at 1000 × g for 5 min. The medium was removed, and the cells were washed three times with 5 ml phosphate-buffered saline. The K-562 cells were extracted with c h l o r o f o r m / m e t h a n o l / w a t e r (10:10:3) twice to remove radiolabeled glycolipids [14]. The cells were further extracted with

255 83% ethanol to remove large polylactosamine-containing glycolipids [15]. The remaining insoluble fraction was extensively digested with pronase [14]. Gel filtration analysis. Glycopeptides obtained following pronase digestion were subjected to gel filtration chromatography on either a column (1.3 x 90 cm) of Bio-Gel P4 or on a mixed-bed column (1.3 x 90 cm) of Bio-Gel P4 (upper 60 cm of column packing) and Bio-Gel P10 (lower 30 cm of column packing). Both materials were minus 400 mesh. The mixed-bed column is referred to as a Bio-Gel P4/P10 column. Columns were eluted with 0.2 M ammonium formate containing 0.05% NaN 3 at a flow rate of 4.0 ml/h. 1on-exchange chromatography. Glycopeptides eluted in the void fraction were desalted and subjected to chromatography on a column of DEAE-Sephacel (1.3 × 10 cm). The column was equilibrated with 50 mM Tris-HC1 (pH 6.5) and the sample was loaded in this buffer. The column was sequentially eluted with 50 mM Tris buffer and with the same buffer containing 0.2 M and then 2.0 M NaC1. The glycopeptides that eluted in the void fraction from the column were pooled and are referred to as the neutral fraction. Lectin-affinity chromatography. Glycopeptides were subjected to affinity chromatography on a small column (1.3 x 1.0 cm) of wheat germ agglutinin-Sepharose 6 MB. The glycopeptides were dissolved in 6 ml phosphate-buffered saline and applied to the column at a flow rate of 0.25 ml/min. The column was sequentially eluted with 34 ml phosphate-buffered saline, 30 ml phosphatebuffered saline containing 0.1 M GlcNAc, and 30 ml phosphate-buffered saline containing 1.0 M GlcNAc at a flow rate of 0.5 ml/min. Chemical treatments. Glycans were mild-acidhydrolyzed with 0.05 M trifluoracetic acid for 10 min at 100°C. Mild periodate oxidation (Smith degradation) was peformed as described by Spiro [16]. Hydrazinolysis and re-N-acetylation were carried out as described by Fukuda et al. [17]. Endo-fl-galactosidase digestion. Various fractions eluted with 1.0 M GlcNAc buffer were treated with endo-fl-galactosidase under conditions described by Fukuda and Matsumura [18], except that the glycopeptides were hydrolyzed for 48 h and 0.02% NaN 3 was included. Band 3 and glycophorin glycopeptide. Band 3

glycoprotein was purified from type O adult blood and from fetal blood by the method of Tsuji et al. [5]. Identification of band 3-containing fractions was confirmed by SDS-polyacrylamide gel electrophoresis as described by Laemmli [19]. Approx. 0.5-1.0 mg band 3 glycoprotein was exhaustively digested with pronase (2 mg in 0.4 ml 5 mM CAC12/25 mM Tris-HC1 (pH 8.0) for 24 h at 50°C, followed by similar additions of enzyme in 0.2 ml at 24 and 48 h) for a total of 72 h under a toluene atmosphere. The enzyme was inactivated by heating at 100°C for 5 rain. Insoluble material was removed by centrifugation and the glycopeptides were separated from other contaminants on a P2 column (1 x 10 cm) eluted with deionized H20. Glycopeptides excluded by the column were labeled by the galactose oxidase/NaB3H4 procedure of Gahmberg and Hakomori [20] as modified by Fukuda et al. [1]. The glycopeptides were subjected to gel filtration chromatography on a BioGel P4/P10 column. Glycopeptides excluded by this second column (V0) were pooled and subjected to chromatography on a column of Bio-Gel P30 (1.3 x 55 cm) eluted under similar conditions. Samples were counted in an LKB Model 1212 liquid scintillation counter after addition of Liquiscint (National Diagnostics, Somerville, New Jersey). Glycophorin-like glycopepfides were isolated from glucosamine-labeled K-562 cells. Pronase digestion products were fractionated by affinity chromatography on immobilized wheat germ agglutinin. The fraction that was retained in the presence of phosphate-buffered saline and eluted by 1 M GlcNAc was then fracuonated by ion-exchange chromatography on QAE-Sephadex as described by Varki and Kornfeld [21]. The fraction that eluted with five or more negative charges was used as the glycophorin-like glycopeptides. These glycopeptides could be depolymerized by mild alkaline hydrolysis, and they were not retained by immobilized wheat germ after desialylation or mild alkaline hydrolysis. t

.

Results

The ability of wheat germ agglutinin to interact with GlcNAc polymers has long been recognized [22-24]. Our observation that adult and fetal

256

erythroglycans (repeating copolymers of GlcNAc and galactose) are also accommodated by wheat germ agglutinin suggested that these various polymers sh~/re structural features. A relative high-affinity interaction between polylactosamines and immobilized wheat germ agglutinin has been identified [12,13]; however, further investigation of this interaction was hampered by the fact that the erythroglycans used in our study were extremely large and complex, with both fetal and adult erythroglycans occurring as mixtures of branched and linear molecules. We have therefore continued our investigations using the glycans expressed by human K-562 erythroleukemic cells. These glycans are exclusively linear and are smaller than their counterparts from adult and fetal erythrocytes.

Composition of the glycan fraction retained with high affinity Metabolically labeled glycopeptides were prepared from K-562 cells after the cells were cultured in medium containing [3H]galactose or [3H]glucosamine. The glycopeptides were prepared by exhaustive digestion with pronase of the material prepared by Folch extraction. The glycopeptides were then fractionated by gel filtration on Bio-Gel P4/P10. A considerable percentage of the total radioactivity (galactose- or glucosaminederived) was eluted at or near the void volume of the column (complex-type glycans having four sialyllactosamine branches are included in this column). This large size was not due to incomplete pronase digestion, as the size distributions were not affected by hydrazinolysis. This fraction is comparable to the fraction from Sephadex G-50 identified by Turco et al. [3] as containing polylactosaminoglycans. These large glycopeptides were pooled and subjected to ion-exchange chromatogrtaphy on a column of DEAE-Sephacel to remove charged glycopeptides. The ability of immobilized wheat germ agglutinin to interact with large, neutral glycopeptides was then tested. As shown in Fig. 1, approx. 70% of the labeled glycopeptides were eluted from the column with buffer alone. Onethird of the material in this direct eluate displayed a weak interaction with the column. This trail was not observed when sucrose was eluted or when glycans were eluted from underivatized Affigel 10.

1<2)

l0 8

E

j

OqM GIcNAc IA)

0

.1MGIcNAc

I

I

M GIcNAc

O-

~B) 1 A]O

M GtcNAc ~S

6 D~ +-

5

4-

44

4

o 73 r0

20 ,~0 60 80

20 40 60 80

EIuiion Volume (mL)

Fig. 1. Wheat germ agglutinin chromatography of the large neutral glycopeptides from K-562 cells. The void fraction obtained by gel filtration on Bio-Gel P4/P10 columns was subjected to DEAE-Sephacel chromatography, and the weakly interacting glycopeptides (i.e., eluted in 50 raM Tris-HC1 (pH 6.5)) were pooled, equilibrated with phosphate-buffered saline, and applied to a column of immobilized wheat germ agglutinin. The column was washed with the same buffer, and bound glycopeptides were eluted with buffers containing 0.1 M GIcNAc, then 1.0 M GIcNAc, as indicated by the arrows. A, galactose-labaled glycopeptides; B, glucosamine-labeled glycopeptides.

The material actually retained by the column was evenly divided between material that was eluted with 0.1 M GIcNAc and material that was eluted with 1.0 M GIcNAc. This high-affinity interaction for large, neutral glycans contrasts with previous studies that demonstrated that small, neutral glycopeptides are not retained by immobilized wheat germ agglutinin [25]. The presence of polylactosamine-containing glycans in the high-affin!ty fraction (1.0 M GlcNAc eluate) was investigated by examining the sensitivity of these glycans to digestion with endofl-galactosidase. This enzyme has been demonstrated by Fukuda and Matsumura [18] to be specific for linear regions of polylactosaminoglycans, although branched glycans have also been suggested to be sensitive to hydrolysis during prolonged digestions. All of the glycopeptides in this high-affinity fraction were extensively digested by this enzyme. There was a large reduction in size as examined by gel filtration on Bit-Gel P30 columns, and very little material was excluded even from Bit-Gel P4 columns (Fig. 2). There were

257

I0

8

TABLE I

7

THE EFFECT OF MILD ACID AND PERIODATE OXIDATION UPON THE INTERACTION OF K-562 GLYCANS WITH WHEAT GERM AGGLUTININ

× J

E E .gu

6

\

5 4

4

44U rQ 0 © Pr"

3 2 1

!l

: \ i

30

40

50 Elution

60

70

i

8O

Volume (mL)

Fig. 2. Endo-~8-galactosidase digestion of the high-affinity fraction obtained by immobilized wheat germ agglutinin chromatography. Galactose-labeled glycopeptides eluted from wheat germ agglutinin-Sepharose with buffer containing 1.0 M GleNAc were digested with endo-B-galactosidase. The glycans were subjected to gel filtration analysis on Bio-Gel P4 before digestion (O . . . . . . O) and after digestion ( e ~ e ) . The void volume was fraction 29. The peak eintion fractions of standards were MangGIcNAc , fraction 45; ManzGlcNAc2, fraction 57; GalGIcNAc, fraction 70.

several discrete, low-molecular-weight peaks. The most prominent peak coeluted with the disaccharide N-acetyllactosamine (LacNAc). Role of nonreducing terminal sugars in wheat germ agglutinin-glycan interactions The experiments with neutral glycans demonstrated that sialic acids are not essential for the high-affinity ]ectin-glycan interaction that can occur between isolated polylactosamines and immobilized wheat germ agglutinin. The following experiments were performed to examine whether other glycans could bind with similar affinity. Although we did not rigorously reexamine a broad spectrum of standard carbohydrates individually for the ability of each glycan to be retained by immobilized lectin, we did examine the behavior of the glycopeptides not eluted in the void region of Bio-Gel P4/P10. This fraction contained a wide variety of complex-type glycans that differed in their degree of branching and in their levels of sialylation. As reported in Table I, none of the glycopeptides in this fraction was retained by the immobilized lectin.

The interaction of immobilized wheat germ agglutinin with small and large glycopeptides from K-562 cells, and the effect of mild-acid hydrolysis and periodate oxidation upon the interaction with the large glycopeptides. The amounts of radioactive glycopeptides, in cpm, that were duted either directly from immobilized wheat germ agglutinin or after 0.1 M or 1.0 M GlcNAc are listed in the columns below. The materials in the void fraction and the included fraction during, gel filtration on a mixed bed of Bio-Gel P4/P10 were separately pooled. The nonvoided material was applied directly to the immobilized wheat germ agglutinin and is referred to as untreated small glycopeptides. The void fraction, referred to as the large glycopeptides, was divided into three samples. One sample was applied directly to the immobilized wheat germ agglutinin, another sample was subjected to mild-acid hydrolysis, as described in the Materials and Methods section, prior to affinity chromatography, and the third sample was subjected to Smith degradation, as described by Spiro [16], before affinity chromatography. Prior to lectin affinity chromatography the glycopeptides were equilibrated with phosphate-buffered saline by means of gel filtration on short Bio-Gel P2 columns and were applied to the immobilized lectin in phosphate-buffered saline. The column was washed with phosphate-buffered saline and the bound glycopeptides were eluted with buffer containing 0.1 M and 1.0 M GlcNAc. Coneentration of GIcNAc

Untreated glycopeptides

Pretreated large glycopeptides

small

large

mild acid

periodate

115260 740 240

86016 13991 3621 ~

105114 2376 4 383

96159 2868 6 874

(M) 0 0.1 1.0

Largeglycopeptides that had not been prefractionated by ion-exchange chromatography were also subjected to lectin-affinity chromatography. These results are also shown in Table I. The high-affinity fraction, which now included acidic glycopeptides, was similar to the corresponding neutral fraction in both size and endo-p-galactosidase sensitivity. Both mild-acid treatment and periodate oxidation (Table I) of the large glycopeptide pool substantially reduced the amount of glycans with moderate affinity (eluted with 0.1 M GlcNAc buffer) but actually increased the percentage of radioactivity eluting in the high-affinity

258 fraction. Again, these high-affinity fractions were sensitive to endo-fl-galactosidase digestion. Thus, polylactosaminoglycan species interacted at high-affinity with immobilized wheat germ agglutinin, and the results suggest that this interaction was not dependent upon the presence of particular nonreducing terminal sugars. The following experiments examined what determinants were in fact required for this high-affinity interaction with immobilized wheat germ agglutinin.

rO

'

0 X

11

E 10 EL 0

10

4-

~= 5 0

it~

, i

20

Structural requirements for high-affinity glycanlectin interactions Glycopeptides metabolically labeled with either galactose or glucosamine were fractionated by size, and the glycans excluded from Bio-Gel P10 columns were then fractionated by their affinity for immobilized wheat germ agglutinin. With glycans that had not been prefractionated by charge, about 13% of the glycans were retained by the immobilized lectin; 75% of these glycans were released by 0.1 M GIcNAc and 25% were released w h e n the GIcNAc concentration was raised to 1.0 M (Fig. 3). These fractions, referred to as moderate- and high-affinity fractions, respectively, were then fractionated by ion-exchange chromatography (Fig. 4). Elution of the weakly anionic glycopeptides was achieved with 50 mM Tris (pH 6.5). The anionic glycopeptides retained in the presence of 50 mM salt were displaced with either 0.2 M or 2.0 M NaCl-containing buffer. Most of the glycopeptides not retained by immobilized wheat germ agglutinin were eluted from DEAE-Sephacel with 50 mM salt. The same is true of the glycopeptides retained by immobilized wheat germ agglutinin in the presence of 0.1 mM GIcNAc. In contrast, the glycopeptides displaced from immobilized wheat germ agglutinin by 0.1 M GlcNAc were either retained by the ion exchange column or substantially retarded during elution with 50 mM Tris (pH 6.5). These various fractions were then investigated as follows. Our initial examination was directed toward the possible role of accessible GlcNAc residues in the glycan-lectin interaction. Our intention was to remove sialic acid and fucose determinants and examine the effect of exposing GlcNAc by subsequent digestion with /3-galactosidase. However, mild-acid treatment of the weakly anionic glyco-

(B) 15

(A)

15

50

80

L

20

'i,,

50

80

Elu'rion Volume (ml) Fig. 3. Wheat germ agglutinin affinity chromatography of large K-562 glycopcptides. The void fraction obtained by gel filtration on Bio-Gel P/4/P10 columns was equilibrated with phosphate-buffered saline and subjected to chromatography on immobilized wheat germ agglutinin. The column was washed with phosphate-buffered saline and the bound $1ycopcptidcs were eluted with 0.1 M GlcNAc buffer (yielding moderate-affinity glycopcptides) and then ¢luted with 1.0 M GIcNAc-buffer (yielding high-affinity glycopcptides). A, galactos¢-labeled glycopeptidcs; B, glucosamine-labeled glycopcptides.

peptides, previously fractionated by affinity chromatography on immobilized wheat germ agglutinin, resulted in substantial shifts in the apparent affinity of many glycopeptides. The glycopeptide population displaced by 0.1 M GlcNAc .-25

(A)

(C)" 25

(B)"

0

20

;l

fIl L

u_ 15 c_

15 lO

el_

E o

5

5 20

40

60 Elution

20 4 0

60

20

40

60

Volume Iml)

Fig. 4. DEAE-Sephacelchromatographyof [3H]ghcosaminelabeled fractions obtained by lectin affinity chromatography. Nonretained (panel A), moderate-affinity(panel B) and highaffinity (panel C) glycopeptidesobtained by wheat germ agglutinin chromatographywere subjected to ion-exchangechromatography. Weakly anionic glycopeptides were obtained by eluting the DEAE-Sephac¢Icolumn with 50 mM Tris-HCl(pH 6.5), while anionic glycopeptideswere removed with the same buffer plus 0.2 then 2.0 M NaCI, as indicated by the arrows.

259 underwent the most dramatic change, in which only 22% of the applied label still eluted in the same concentration of GlcNAc, and the remaining material was evenly divided between material that was now either not retained or retained even in the presence of 0.1 M GlcNAc (Fig. 5B). This loss of affinity for the immobilized lectin was not due to extensive degradation of the cabohydrates, since the galactose-labeled products of this treatment were still eluted in the void volume of Bio-Gel P 4 / P 1 0 columns. Mild-acid hydrolysis of glycopeptides that had a high-affinity interaction with immobilized wheat germ agglutinin resulted in a smaller proportional shift, so that now 25% was not retained and 5% was eluted with 0.1 M GlcNAc buffer (Fig. 5C). (The remaining 70% of the applied label was still retained in the presence of 0.1 M GlcNAc and eluted with 1.0 M GlcNAc). Mild-acid treatment of the fraction that showed no interaction with immobilized wheat germ agglutinin resulted in a relatively small proportional shift in binding affinity (Fig. 5A). However, the 4% shifted into the high-affinity fraction is equivalent in amount to the original high-affinity fraction obtained from untreated material (Fig. 3).

Contribution of accessible GIcNAc to glycan-lectin interaction We have used the exoenzymes B-hexosaminidase and B-galactosidase to assess the contribution that terminal GlcNAc residues play in the glycan-wheat germ agglutinin interaction. These enzymes are active on these glycopeptides and in combination they result in the depolymerization of mild-acid-treated glycopeptides. We have examined the requirement of terminal GlcNAc exposure for the high-affinity glycan-lectin interaction using fl-hexosaminidase. Treatment of glycopeptides, retained by immobilized wheat germ agglutinin in the presence of 0.1 M GIcNAc, with this enzyme did not reduce the apparent affinity of the lectin-glycan interaction. This is true for glycopeptides that had high affinity before mild acid-treatment (Fig. 5E) and also for glycopeptides that had enhanced affinity on mild-acid treatment (Fig. 5F). We have also explored the possibility that potential lectin-glycan interactions were prevented

(A)

3 o 2 x 1

rO I

,L

1

(C)

(B)

1 12 1

Lj~-

(D}

4 5

(E)

(F)

o-5 t) 2

4

3 2 1

1 L

20

60

20 Elution

60

20

60

Volume(ml)

Fig. 5. Effect of chemical and enzymatic treatment on wheat germ agglutinin binding affinity of mildly anionic glycopeptides. Large, glucosamine-labeledglycopeptidesthat weakly interacted with DEAE-Sephacei(i.e., eluted in 50 mM Tris-HCI (pH 6.5)) were subjected to wheat germ agglutinin affinity chromatography. The materials not retained and those retained with moderate- and high-affinity were separately pooled and treated with mild acid. Panels A, B, and C show the elution behavior of the mildly anionic fractions previouslynot retained or retained with moderate and high affinity, respectively, on rechromatography after mild-acid hydrolysis. The glycopeptides still not retained by wheat germ agglutinin after mild-acid treatment (panel A), were pooled and subjected to fl-galactosidase treatment and rechromatographed on the lectin affinity column, as shown in panel D. Similarly, fractions from the moderate- and high-affinity fractions that eluted with high affinity after the mild-acid treatment (panels B and C) were pooled, treated with fl-N-acetyibexosaminidase,and subjected to wheat germ agglutinin affinity chromatography as shown in panels E and F, respectively. The first arrow indicates elution with 0.1 M GlcNAc buffer, while the second indicates elution with 1.0 M GIcNAc buffer.

by subterminal GIcNAc being obscured by galactose. Treatment of glycopeptides that were not retained by immobilized wheat germ agglutinin after mild-acid treatment with fl-galactosidase did not result in glycopeptides that were now retained by the lectin. Thus it appears that terminal GlcNAc is not sufficient for lectin-glycan interactions strong enough to result in glycopeptides being retained by immobilized lectin.

Contribution of sialic acid residues to glycan-lectin interaction The binding of anionic glycopeptides (eluted with 0.2 M NaC1/50 m M Tris buffer from DEAE-Sephacel) to immobilized wheat germ ag-

260

glutinin could be due to the presence of sialic acid residues. However, we have found shifts to both higher and lower affinity on desialylation, indicating that some sialic acid residues may be responsible for moderate-affinity interactions, whereas others may actually shield potential high-affinity interactions. Anionic glycopeptides eluted from immobilized wheat germ agglutinin by 0.1 M and 1.0 M GIcNAc were subjected to neuraminidase treatment (0.5 units, pH 5.5, 37°C, 24 h) to remove sialic acid residues. This treatment of the moderate-affinity (0.1 M GlcNAc) fraction with neuraminidase indicated that a considerable amount of the GlcNAc label was present as sialic acid. One-third of the glucosamine label was released as sialic acid. Following desialylation, approx. 75% of the polymer-associated glucosamine label was not retained by immobilized wheat germ agglutinin (Fig. 6A); however, it was now retained by immobilized peanut agglutinin and displaced by 0.1 M lactose. Neuraminidase treatment did not result in obvious fragmentation, as digestion products of galactose-labeled material were still excluded from Bio-Gel P10 columns (Fig. 7, open symbols, dashed lines). However, following mild alkaline hydrolysis, in the presence

CA)

3

[B) 4

rO

'o X

£ 20

40

60

80

Elution

, 20

40

60

80

Volume (ml)

Fig. 6. Effect of neuraminidase treatment on the binding of moderate- and high-affinityanionic glycopeptides to wheat germ agglutinin. Moderate- and high-affinityanionic glycopeptides •luted from DEAE-Sephaeel with 0.2 M NaCI were treated with neuraminidase and the remaining glycopeptides were analyzed by wheat germ agglutinin affinity chromatography. Wheat germ agglutinin elution profiles following enzymatictreatmentof moderate-affinityglycopeptides(panel A) or high-affinityglyeopeptides(panel B).

2 rO i © X

E

;?

EL C)

30

50

Elution

70

90

Volume (ml)

Fig. 7. Gel filtration analysis of neuraminidase-derived products. The nonretained wheat germ agglutinin binding fraction seen in Fig. 6A was subjected to gel filtration on Bio-Gel P4/P10 columns as described in the Materials and Methods section. (O . . . . . . O), profile obtained before mild-alkaline hydrolysis; • • , profile obtained after mild-alkaline hydrolysis for 16 h.

of sodium borohydride, virtually all the radioactivity was included on Bio-Gel P4/P10 column, with a substantial peak co•luting with the disaccharide LacNAc (Fig. 7, dosed symbols, continuous lines). The sensitivity of these glycopeptides to mild alkaline treatment and the decrease in their affinity for immobilized wheat germ agglutinin upon loss of sialic acid directly parallels the reported properties of glycophorin glycopeptides. Only a small portion (9%) of the moderate-affinity fraction was unchanged by desialylation; that is, still eluted from immobilized wheat germ agglutinin by 0.1 M GlcNAc buffer. Surprisingly, desialylation of the anionic glycopeptides displaced from immobilized wheat germ agglutinin by 0.1 M GlcNAc resulted in a population of these glycopeptides (16%) having increased affinity for wheat germ agglutinin and not being displaced by 0.1 M GIcNAc (Fig. 6A). These now high-affinity glycopeptides were extensively depolymerized by endo-fl-galactosidase digestion. The anionic glycopeptides retained by immobilized wheat germ agglutinin in the presence of 0.1 M GlcNAc demonstrated substantially different behavior following neuraminidase treatment. Digestion with this glycosidase released 15% of this glucosamine-labeled fraction as sialic acid. Subsequent wheat germ agglutinin affinity chromatogra-

261 phy of this desialylated fraction revealed that 90% of the polymer-associated label was still retained in the presence of 0.1 M GlcNAc buffer (Fig. 6B). Only 5% of the anionic high-affinity fraction was not retained by immobilized wheat germ agglutinin following neuraminidase treatment.

(A)

12

x 12

10 8

26 Contribution of fucose residues to glycan-lectin interaction The finding that sialylation could obscure potential high affinity binding sites prompted a closer investigation of the role that other decorations might play in making glycan determinants accessible to immobilized lectin. The shifts observed earlier in the binding affinity of weakly anionic glycopeptides following midl-acid hydrolysis could be due to the removal of either sialic acid or fucose residues (Fig. 5A-C). The following experiments demonstrate that the presence of fucose decorations influenced the affinity of glycan-lectin interactions. However, this was not just a simple shielding, as both increases and decreases in affinity were observed after removal of fucose. Weakly anionic glycopeptides displaced from immobilized wheat germ agglutinin by 0.1 M and 1.0 M GIcNAc were subjected to fucosidase (Charonia lampis) digestion. The results are shown in Fig. 8. Examination of these digestion products by gel filtration revealed that no glucosamine or galactose label was released by a-fucosidase treatment. More than half of the moderate-affinity (0.1 M GIcNAc) fraction were not retained after fucosidase treatment, although most of the glycopeptides showed a weak interaction with the lectin and were retarded during elution (Fig. 8A). Fucosidase treatment of weakly anionic high-affinity (1.0 M GIcNAc) glycopeptides resulted in a substantial loss of affinity of the glucosaminelabeled glycopeptides in this fraction (Fig. 8B). Less than half of the total glucosamine-labeled glycans were still retained in the presence of 0.1 M GlcNAc. Of the remaining glycans, half were displaced by 0.1 M GlcNAc buffer, and the other half were merely retarded. No loss of glucosamine label from the glycans was observed on fucosidase digestion, indicating that the altered affinity was not the result of other glycohydrolase reactions. Fucose has not previously been implicated in glycan-wheat germ agglutinin interactions, and we

6

4 2 20

40

60

8O

Elotion

20

40

60

80

Volume (ml)

Fig. 8. Effectof fueosidasetreatmenton glycopeptidesretained by the immobilizedwheat germ agglutinin. Moderate-affinity and high-affinity glucosamine-labeled glycopeptides that weakly interacted with DEAE-Sephacel, i.e., eluted in 50 mM Tris-HCl (pH 6,5), were subjected to fucosidasetreatment (C. lampis) and again analyzed by wheat germ agglutinin chromatography. Elution profiles of moderate-affinity(panel A) and high-affinity (panel B) glycopeptides are shown after fucosidase treatment.

have found that inclusion of 0.1 M fucose in the GlcNAc-containing buffers does not affect the elution behavior of these glycans. This suggests that fucose is not a part of the glycan determinant that interacts directly with wheat germ agglutinin. The increased affinity for wheat germ agglutinin that is observed on the removal of fucose may be a result of fucose obscuring a potential high-affinity site. The decreased affinity for wheat germ agglutinin that is observed with other glycans on the removal of fucose is more difficult to understand. Fucosylation of these glycans may affect their conformation, making potential high-affinity sites more accessible. The role of glycan conformation in establishing high-affinity lectin-glycan interactions was explored by varying the density of the immobilized lectin.

Role of lectin density in promoting high-affinity interactions Wheat germ agglutinin was imobilized at different total protein concentrations varying from 0.1 to 10 mg/ml. Five lectin densities were explored in depth using adult and fetal erythroglycans and the glycophorin-like glycopeptides from K-562

262

cells. The results of these experiments are summarized in Fig. 9, which shows the proportion of each glycan in the nonretained and retarded classes and the two retained classes that were eluted with either 0.1 M or 1.0 M GlcNAc. The three glycans showed very similar profiles for the decrease in nonretained material (Fig. 9A, E and I), for the decrease in the retarded material (Fig. 9B, F and J) and for the increase in total retained material (Fig. 9C; Fig. 9G and K, dotted lines). However, although fetal and adult erythroglycans showed increasing amounts of material not displaced by 0.1 M GlcNAc (Fig. 9H and L), glycophorin-like glycopeptides never showed this high-affinity interaction (Fig. 9D). This wide spectrum of interactions for each glycan type suggests that the highaffinity interaction is not the result of an intrinsically high-affinity sequence but more likely the result of the large glycans being able to span -o

0'8 x

"'

06 ~ 04 0,2

T3._

E

0.8

t

o.z

(A)

(B)

.

(D}

0"8

/' ,~,,," ,,~,,,

~

m O6~ ~n 0,4

(0

(E)

0.6 0.4 0.2

(G) ..~ IH) o,, . , ~

(F)

0.8 0.6 0.4 o.z

,

co u

\ 0.8 ~ 045

m

0"4

"

0'2

(I)

tj)

{K)

,

,

.~.-.~, ,~(L].

0.8 0.6 0'4

/ ~"

. . . ~ 06

50

Concentration

0"2 05

3<) 06

3'0 05

of' Immobilized W G A

50 (mg/ml)

Fig. 9. Effect of wheat germ agglutinin concentration upon the interaction of glycans with immobilized wheat germ agglutinin. Wheat germ agglutinin was immobilized on Affigel 10 at 0.1, 0.3, 1, 3 and 10 mg/ml. The acidic, moderate-affinity glycans from K-562 cells (A-D) and the glycopcptides prepared from hand 3 glycoproteins from adult (E-H) and fetal (I-L) erythrocytes were chromatographed on columns containing wheat germ agglutinin immobilized at these different concentrations. The proportions of the applied material that eluted directly in phosphate-buffered saline (A, E and I), that were retarded but duted in phosphate-buffered saline (B, F and J), and that were eluted in 0.1 M (C, G and K) and 1.0 M (D, H and L) GIcNAc are shown as a function of the concentration of immobilized wheat germ agglutinin. The proportions of each glycan fraction that were retained by each column (i.e., not ¢luted by phosphate-buffered saline) are also shown in C, G and K (O . . . . . . O).

adjacent lectin molecules. The proportion of the glycans that span adjacent molecules is a function of both the proximity of the lectin molecules (determined by the protein density) and the effective glycan size (determined by the glycan conformation). Discussion

Plant lectins have been widely used for many years as probes for carbohydrate structures on animal cell surfaces [26-30]. Recently, lectin affinity chromatography has reestablished itself as a powerful tool for isolating glycopeptides with specific carbohydrate binding determinants [31]. However, it is apparent that the present knowledge of structures that interact with even the most commonly used immobilized lectins is incomplete. We found that linear polylactosamines were specifically retained by wheat germ agglutininSepharose, and have examined the role that glycan conformation plays in this binding interaction. Few studies have identified an interaction between wheat germ agglutinin and defined glycopeptides. Yamamoto et al. [25] reported that immobilized wheat germ agglutinin would not specifically retain any high-mannose, hybrid, or complex N-linked glycopeptides that they tested. However, hybrid-type glycopeptides were found to interact sufficiently to be retarded during elution from immobilized wheat germ agglutinin. Debray et al. [32] examined the ability of oligosaccharidic structures to inhibit hemagglutinin by wheat germ agglutinin. Out of a broad panel of glycoproteins and glycopeptides, the only oligosaccharidic structure that was found to inhibit hemagglutinin was the glycopeptide derived from ovomucoid. This glycopeptide contains, in addition to chitobiose, sequences proximal to asparagine, five exposed nonreducing N-acetylglucosamine residues. Polylactosamines that contain alternating galactose and GIcNAc residues have been shown to interact with wheat germ agglutinin. Carlsson et al. [33] demonstrated that desulfated keratan sulfate formed a precipitate with wheat germ agglutinin and Gallagher et al. [12] and Ivatt et al. [13] have demonstrated an interaction between polylactosamines and immobilized wheat germ agglutinin. This interaction is distinct from that demonstrated by

263 Bhavanandan et al. [34], which is sialic-acid dependent. Bhavanandan et al. [34] demonstrated that the binding of an anionic glycopeptide fraction to immobilized wheat germ agglutinin could be competed with 0.1 M GlcNAc buffer. This glycopeptide was composed of several adjacent O-linked glycans. Removal of more than 50% of its sialic acid resulted in loss of wheat germ agglutinin binding. Alternatively, mild-alkaline hydrolysis, which resulted in conversion to individual tri- and tetrasaccharides, also resulted in loss of wheat germ agglutinin binding. The glycopeptides did not contain GIcNAc, but did possess galactose-GalNAc linkages. Anionic glycopeptides with similar physical and lectin-binding properties were identified in the present study. Neuraminidase treatment resulted in a loss of 30% to 35% of the glucosamine label as sialic acid. This desialylation was accompanied by a 75% decrease in lectin binding. The fraction not retained after desialylation was susceptible to mild alkaline hydrolysis, with the major product a disaccharide that migrated with LacNAc. An absolute requirement for terminal sialic acid or accessible fl-linked GlcNAc for wheat germ agglutinin binding is not supported by the present data. Glycopeptides that lost their affinity for the immobilized wheat germ agglutinin following neuraminidase treatment were determined to be nonpolylactosamine glycopeptides. By contrast, glycopeptides that increased their affinity for the immobilized lectin following neuraminidase treatment were polylactosamines, based upon their sensitivity to endo-fl-galactosidase. Removal of fllinked GlcNAc resulted in a small loss of GlcNAc label but did not cause any detectable change in affinity for the lectin column. It was unexpected that fucosidase treatment would have such a great effect on wheat germ agglutinin binding, because fucose is not recognized as part of the carbohydrate binding determinant for this lectin. In addition to al,6 linkage to the proximal GIcNAc of N-linked oligosaccharides, fucose residues are also found on polylactosamines linked al,2 to galactose and al,3 to GlcNAc. A mouse melanoma line that demonstrated increased resistance to wheat germ agglutinin toxicity expressed a 60-70-fold increase in al,3-fucosyltransferase compared to the wild

type [35]. This additional fucose was suggested to be the biochemical lesion directly responsible for the resistance to wheat germ agglutinin. Whether this lesion is direct shielding or an indirect conformational effect is not known. The information presented in this report suggests that the interaction of polylactosamines with immobilized wheat germ agglutinin is very complex and that although repeating LacNAc units are required for the binding, they are not by themselves sufficient for the interaction to occur. For example, the presence of fucose residues on the glycans appears to be essential for some polylactosaminoglycans to interact with the lectin affinity column. Presumably, the sites for this binding interaction are GlcNAc-fl-l-3-Gal-fll-4GlcNAc-fll-3-Gal units, either terminal or internal ones. Polylactosaminoglycans are often referred to as 'erythroglycan' or 'embryoglycan'. These terms suggest a single species. The present report emphasizes the heterogeneity of the polylactosaminoglycans as a class of carbohydrates. If anything, the present study represents an oversimplification of the existing situation. Polylactosaminoglycans are extremely complex carbohydrates possessing several sites for terminal decorations. Polylactosamines probably differ in the number of LacNAc branches as well as the number of LacNAc units on each branch. The presence of ill-6 branches adds to this complexity [5]. The embryonic glycans currently under investigation in this laboratory appear to be considerably larger and more complex than K-562 or erythrocyte-derived polylactosaminoglycans. Elucidation of the functional role of these carbohydrates requires a more extensive knowledge of their structure. Wheat germ lectin affinity chromatography may prove to be very useful for the future study.

Acknowledgments We would especially like to thank Drs. Michiko and Minoru Fukuda for their generous gifts of both band 3 glycopeptides and endo-fl-galactosidase, and Dr. Toshiaki Osawa for his generous gift of Bacillus fulminans and a-fucosidase. We wish to thank the staff of the Labor and Delivery Unit of Methodist Hospital for their kind efforts in

264

providing us with cord blood samples, and Dr. Christopher Reading of our department for providing the K-562 cells. We thank Nancy Edwards for help in preparing this manuscript. References 1 Fukuda, M., Fukuda, M.N. and Hakomori, S.-I. (1979) J. Biol. Chem. 254, 3700-3703 2 Kapadia, A., Feizi, T. and Evans, M.J. (1981) Exp. Cell. Res. 131, 185-195 3 Turco, S.J., Rush, J.S. and Laine, R.A. (1980) J. Biol. Chem. 255, 3266-3269 4 Ivatt, R.J. (1984) in The Biology of Glycoproteins (Ivatt, R.J., ed.), pp. 95-181, Plenum Press, New York 5 Tsuji, T., Irimura, T. and Osawa, T. (1980) Biochem. J. 187, 677-686 6 Zdebska, E. and Koscielak, J. (1978) Eur. J. Biochem. 91, 517-525 7 Fukuda, M. and Fukuda, M.N. (1981) J. Supramol. Struct. Cell. Biochem. 17, 313-324 8 Marsh, W.L. (1961) Br. J. Haematol. 7, 200-209 9 Jarnefelt, J., Rush, J., Li, Y.-T. and Laine, R.A. (1978) J. Biol. Chem. 253, 8006-8009 10 Krusius, T., Finne, J. and Ranvala, H. (1978) Eur. J. Biocohem. 92, 289-300 11 Fukuda, M., Dell, A. and Fukuda, M.N. (1984) J. Biol. Chem. 259, 4782-4791 12 Gallagher, J.T., Morris, A. and Dexter, T.M. (1985) Biochem. J. 231, 115-122 13 Ivatt, R.J., Harnett, P.B. and Reeder, J.W. (1986) Biochim. Biophys. Acta 881, 124-134 14 Turco, S.J., Stetson, B. and Robbins, P.W. (1977) Proc. Natl. Acad. Sci. USA 74, 4411-44"14 15 Hakomori, S.-I. (1978) Methods Enzymol. 50, 207-211

16 Spiro, R.G. (1972) Methods Enzymol. 28, 3-42 17 Fukuda, M., Kondo, T. and Osawa, T. 1(1976) J. Biochem. 80, 1223-1232 18 Fukuda, M.N. and Matsumura, G. (1976) J. Biol. Chem. 251, 6218-6225 19 Laemmli, U.K. (1970) Nature 227, 680-685 20 Gahmberg, C.G. and Hakomori, S.I. (1973) J. Biol. Chem. 248, 4311-4317 21 Varki, A. and Kornfeid, S. (1981) J. Biol. Chem. 256, 9937-9943 22 Allen, A.K., Neuberger, A. and Sharon, N. (1973) Biochem. J. 131, 155-162 23 Goldstein, l.J., Hammarstrom, S. and Sundblad, G. (1975) Biochim. Biophys. Acta 405, 53-61 24 Monsigny, M., Delmotte, F. and Helene, C. (1978) Proc. Natl. Acad. Sci. USA 75, 1324-1328 25 Yamamoto, K., Tsuji, T., Matsumato, I. and Osawa, T. (1981) Biochemistry 20, 5894-5899 26 Goldstein, I.J. and Etzler, G.E. (1983) Prog. Clin. Biol. Res. 138, 1-314 27 Goldstein, I.J. and Hayes, C.E. (1978) Adv. Carbohydr. Chem. 35, 127-340 28 Lotan, R. and Nicolson, G.L. (1979) Biochim. Biophys. Acta 559, 329-376 29 Roth, J. (1978) Exp. Pathol. S3, 1-186 30 Briles, E.B. (1982) Int. Rev. Cytol. 75, 101-165 31 Cummings, R.D. and Kornfeld, S. (1982) J. Biol. chem. 257, 11235-11240 32 Debray, H., Decout, D., Strecker, G., Spik, G. and Montreuil, J. (1981) Eur. J. Biochem. 117, 41-55 33 Carlsson, H.E., Lonngren, J., Goldstein, I.J., Christner, J.E. and Jourdian, G.W. (1976) FEBS Lett. 62, 38-40 34 Bhavanandan, V.P., Umemoto, J., Banks, J.R. and Davidson, E.A. (1977) Biochemistry 16, 4426-4437 35 Finne, J., Burger, M.M. and Prieels, J.-P. (1982) J. Cell Biol. 92, 277-282