Isolation and Characterization of a Glycoprotein from Human Group 0 Erythrocyte Membrane*

Abstract A glycoprotein was isolated from human group O erythrocyte membranes previously treated with detergents by a mild extraction with 90% phenol and further purified by successive passages through columns of DEAE-, CM-Sephadex, and Sephadex G-100 followed by a chloroform-methanol extraction. The purified glycoprotein appeared to be a single molecular species in gel filtration, polyacrylamide electrophoresis, and ultracentrifugation, and a molecular weight of 53,000 was estimated by sedimentation equilibrium experiments in the presence of 1% sodium dodecyl sulfate. This glycoprotein, which contains 55% carbohydrate, has H and MN blood group activities and carries the receptors for various phytohemagglutinins. After alkaline, reductive degradation of the glycoprotein, the blood group H and MN activities were greatly diminished, whereas the inhibitory activities against certain phytohemagglutinins, especially concanavalin A, were remarkably enhanced, indicating that the antigenic receptors of the former blood group activities clearly reside in alkali-labile O-glycosidically linked oligosaccharide chains and the receptors for the latter phytohemagglutinins mostly reside in alkali-stable N-glycosidically linked oligosaccharide chains. The purified glycoprotein was also found to be a potent inhibitor for [3H]thymidine incorporation in human lymphocytes stimulated by various phytomitogens and anti-lymphocyte serum.

Many investigators have described the isolation from human erythrocyte stroma of glycoprotcins which have MN blood group specificity and myxovirus receptor activity (l-7).
Recently, certain glycopeptides obtained by the digestion of human erythrocytes with proteolytic enzymes were found to possess inhibitory activity against various phytohrmagglutinins and ABH blood groupspecific antiserum (S-10).
This paper describes the purification and characterization of a glycoprotein from human group 0 erythrocyte membranes which is found to be a potent inhibitor * This research was supported by a grant from the Llinistry of Rducation of Japan.
for various hemagglutinins and also for [311]tllymidine incorporation in human peripheral lymphocytes stimulated by various mitogens.
Also reported in this paper arc t,he changes of hapten inhibitory activity of this glycoprotein after alkaline, reductive degradation.
Based on these results, the nature of the antigenic receptors on the human erythrocyte surface is discussed.

EXPERIMENTAL PROCEDURE
Preparation of Erytkrocyte Stroma-Outdated bank blood of group 0 was kindly provided by Nihon Seiyaku Co., Tokyo. The erythrocyte stroma was prepared from washed cells according to the method described by Yamakawa and Suzuki (11).
Purification of Glycoprotein from Eryihrocyte Stroma-Thirty grams of lyophilized stroma were suspended in 1500 ml of 0.14 M NaCI-0.02 M phosphate buffer, pH 7.0, and the suspension was stirred for 2 hours. Bfter centrifugation at 7000 rpm for 15 min, the residual stroma was suspended in 700 ml of the same buffered saline conhining lcjO Triton X-100 and 0.5% sodium dodecyl sulfate, and the suspension was stirred for 24 hours at 4". Then, 650 ml of cold 90% phenol was added and the mixture was vigorously stirred for 2 hours. After centrifugation, the aqueous layer was dialyzed against tap water and then centrifuged.
The clear supernatant obtained was designated as crude glycoprotein. Further purification of the crude glycoprotein was achieved by successive chromatographies on DEAE-Sephadex A-25, CM-Sephadex C-50, and Sephadex G-100, as follows.
The crude glycoprotein was dialyzed against 0.05 &I-NaCl-0.02 M phosphate buffer, pH 6.0, containing 0.4% Triton X-100 and 5 mM EDTA, and applied to a column of DEAE-Sephadex A-25 equilibrated with the same buffer, and gradient elution was performed as shown in Fig. 1. The fractions containing the major glycoprotein were pooled.
Since removal of the detergents could be greatly facilitated by dialysis in the presence of ethanol, ethanol was added to 50% to the pooled fraction and the mixture was dialyzed against 0.02 M acetate buffer, pH 4.7. The dialysate was then applied to a column of CM-Sephadex C-50 (5.5 x 14 cm) equilibrated with the same buffer, and elution was carried out with the same buffer.
Most of the applied glycoprotein did not bind to the CM-Sephadex column under these conditions, and separation from the contaminating proteins, which bound to the column, was effectively achieved by this procedure.
To the glycoprotein fraction was added ethanol to 50%, and the solution was exhaustively dialyzed against water to remove the detergents, concentrat'ed by means of ultrafiltration, and applied to a column of Sephadex G-100. The glyeoprotein was eluted as a single peak. After lyophilization, the glycoprotcin was dissolved in water to a concentration of 50 mg per ml, and the solution was cstracted with 20 volumes of chloroform-methanol (2 : 1, v/v), dialyzed against distilled water and lyophilized.
Over-all yield of the purified glycoprotein was 790 mg from 100 g of lyophilized stroma.
Phytohemagglutinins and Antisera-Phaseolus vulgaris hemagglutinin was a product of Difco.
Other phytohcmagglutinins were cstracted from the rcspcctive seeds by the method previously described (22). Pokewced mitogen was purified by the method of Reisfeld et al. (23). Anti-human lymphocyte horse serum was kindly provided by Dr. hlatsukura, Institute of Medical Science, University of Tokyo.
The absorbed rabbit anti-M and -N immune sera were prepared according to the method described by Race and Sangcr (24).
Ilemagglutination and Ilemagglutination Inhibition Assays-The titration and inhibition assays using human crythrocytes freshly obtained from a donor were carried out according to the methods previously described (15,19).
Lymphocyte Cultures for Mitogenic Assay-Human peripheral lymphocytes were cultured by the method previously described by Toyoshima et al. (15). Inhibition assays for [6-3H]thymidine incorporation into the lymphocytes with sugars and glycopeptides were performed as described previously (21). The inhibition assays were usually carried out four times on each inhibitor against the same mitogen, and an average value of the incorporation was calculated.
Jlolecular Weight Determination-The molecular weight of the purified glycoprotein was measured by sedimentation equilibrium \Tith a Hitachi model UCA-IA ultracentrifuge according to the method of Y&antis (25). The concentrations of the glycoprotein t,ested were 0.33, 0.5, and 10/O in 0.01 M phosphate buffer, pH 7.0, containing 1 To sodium dodecyl sulfate.
Disc Electrophoresis-Disc electrophoresis in polyacrylamide gels was carried out according to Weber and Osborn (26). Thirty to one hundred micrograms of the purified glycoprotein was placed onto a 7.5y0 gel in 0.01 M phosphate buffer, pH 7.0, containing 0.1 y. sodium dodecyl sulfate.
Staining for protein was performed with Amido black in 7% acetic acid, and carbohydrate Teas detected by the periodate-Schiff method of Zacharius et al. (27).
Carbohydrate Determinations-Total neutral sugar was determined by the phenol method of Dubois et al. (28). Total sialic acid was determined by the periodate-resorcinol method of Jourdian et al. (29), using N-acetylneuraminic acid as a standard. Total hexosamines and the individual amounts of glucosamine, galactosamine, and galactosaminitol were obtained, after hydrolysis with 4 N HCl for 6 hours at loo", on a Hitachi KLA-313 amino acid analyzer.
A 50.cm column of Amberlite IR-120 was used, and clution was carried out with 0.35 N sodium citrate buffer, pH 5.28. Quantitative determination of individual neutral sugars was carried out b, 7 gas-liquid chromatography after reduction to the respective alditol, followed by trifluoroacetylation according to the method of Matsui et al. (30), as described b) >latsumoto and Osawa (20) and Akiyama and Osan-a (10).
Samples for this assay were hydrolyzed with 1 N I-I&0( at 100" for 8 hours. Amino dcid Determination-Amino acids were determined after hydrolysis for 24 hours in constant boiling IICl (5.7 N) at 107" in sealed, evacuated tubes using a Hitachi KLA-3J) amino acid analyzer.
Tryptophan was determined on unhydrolyzcd protein samples by the spectrophotomctric method of Goodwin and Morton (31).
dlkaline Treatment of Erythrocyfe G&oprotein-Glycoprotcin was dissolved in 0.2 M r\l'aOII-0.4 M -\iaBI& and the solution was incubated for 24 h or 48 hours at room tcmperaturc in a sealed tube, in the dark, under nitrogen (32). The excess borohydridc n-as destroyed by the careful addition of 2 s acetic acid to $1 6.0, and an aliquot of the mixture n-as applied to a column of Bio-Gel P-6. The void volume fractions of &o-Gel P-6 eluate, containing glycoprotein enriched in alkali-resistant N-glycosididicallg linked oligosaccharidcs, were pooled, dialyzed, and lyophilized.

Purification
of Crude Glycoprotein-A glycoprotein was estracted from freeze-dried ghost membranes after treatment with detergents.
The crude glycoprotcin thus obtained \vas applied to a column of DEAE-Sephades A-25 and the elution of the glycoproteins \vas effected only in the presence of Triton X-100, as shown in Fig. 1. The major glycoprotein fractions were pooled and then applied to a column of CM-Sephadex C-50. This procedure was most effcctivc in removing contaminating proteins. The glycoprotein was finally purified by gel filtration on Sephadex G-100.
In order to remove a tract of contaminating glycolipids the glycoprotein was extracted with chloroform-methanol under the conditions described by Folch et al. (34). The purified glycoprotein was subjected to gel filtration on Sephades G-100 in the presence of 1 7. sodium dodecyl sulfate, and a single peak was observed as shown in Fig. 2. The homogeneity of the purified glycoprotein was further confirmed by ultracentrifugation and disc electrophorrsis. 1. DEAE-Sephadex chromatography of crude glycoprotein. Three hundred milligrams of crude glycoprotein were dialyzed against 0.02 M phosphate buffer, pI1 6.0, containing 0.05 M NaCl, 0.47; Triton X-100, and 5 mM P:DTA (starting buffer), and applied to a DEAIMephadex A-25 column (4.3 X 22.5 cm) equilibrated with the same buffer. The column was washed with 100 ml of the starting buffer, and gradient elution was then performed with 600 ml of the starting buffer in the mixing vessel and 600 ml of 0.02 M phosphate buffer, pI1 6.0, containing 0.6 M NaCl, 0.4% Triton X-100, and 5 rnM EDTA in the reservoir.
Fractions of 16 ml were collected at a flow rate of 30 ml per hour. Fractions were analyzed for sialic acid by the pcriodate-resorcinol method (29) (35). The reduced and alkylated glycoprotein (12 mg) in 3 ml of 1% sodium dodecyl sulfate was applied to a 1.5 X 90 cm Sephadex G-100 column equilibrated with 1% sodium dodecyl sulfate. with the same solution.
Elution was performed Fractions of 2.5 ml were collected at a flow rate of 10 ml per hour. Fractions were analyzed for sialic acid by the periodate-resorcinol method (29)   Analytical Results-Ultracentrifugation of the purified glycoprotein yielded a single symmetrical peak during the whole of the run. A molecular weight of 53,000 was estimated for the purified glycoprotein in the presence of 1 y. sodium dodecyl sulfate by means of sedimentation equilibrium at a speed of 7640 rpm assuming a partial specific volume of 0.66 which was calculated from the analytical data (36,37).
The homogeneity of the purified glycoprotein was further confirmed by disc electrophoresis on polyacrylamide gel in the presence of 0.1% sodium dodecyl sulfate. A single band was obtained at pH 7.0, as shown in Fig. 2B.
The amino acid composition of the purified glycoprotein is presented in Table I. The most notable feature of the amino acid composition of the glycoprotein is the high proportion of serine, threonine, and glutamic acid and the absence of cysteine and tryptophan.
Fifty-five per cent of the dry weight of the purified glycoprotein is carbohydrate. Carbohydrate analyses on the purified glycoprotein revealed that sialic acid, galactose, and galactosamine were the predominant sugars, with smaller amounts of glucosamine, mannose, and fucose. Quantitative data from the foregoing carbohydrate analyses are included in Table I.
In order to ensure the absence of glycolipids in the purified glycoprotein, the determination of sphingosine bases was performed by means of gas-liquid chromatography after methanolysis followed by trifluoroacetylation (38) using galactocerebroside from bovine brain as a reference compound. No trace of sphingosine bases was detected.
Hemagglutination Inhibitory Activity of Purified Glycoprotein- The prlrifietl glycoprotein (30 mg in A, 20 mg in 11 was treated with alkaline borohydride as described in the test for 24 hours (A) or 48 ho\lrs (B), and then applied to a Bio-Gel P-6 column (l.(i X 48 cm). I~:lution was performed with distilled water and fractions of 3 ml wcrc collected at a flow rate of 6 ml per hour. Fractions were :~11:11yzcd for sialic acid by the pcriodnt,c-resorcinol m&hod ('L!)) (0 0) and for hexose by the pllcllol-sulfuric acid n~thod (28) (O-m-~-0 ).
various ~!l~~toller~iag~l~~t illills. Ihrt h~~rmor~~, the glycoprotein has I I :urtl 11x blood ~roul) activilics PYCII after the chloroformmcthallol extraction, ilKlic:atillg thcW l)lood group antigens also rcsidc in the glycoproteill ])art of llllnlilll c>ryt,hrocytc membrane. A lkali~iae-Borohy~ride Treatment o~Pur<~ied Glycoprotein-It has been known that the crytlirocyto ~lycoprotein contains two types of oligosaccharide &ills: one is linked 0-glycosidically to a serinc or a threoninc r&iduc and WI bc sclcctively released from the pcptide backbone 1,~ alkaline t.rc~atmcntZ, while the other is linked .I--glycosidically to a11 aspara$nc residue and is alkalistable. In order to dctc~rlninc which oligosaccharide chain the variolls ant.igcnic receptors rcsidc, the ulkaline, reductive degradatioll of the purified $ycoprotcin n-as carried out and separatio of the modified $ycoprotcin from the released oligosaccharidcs was performed by gel filtration as shown in Fig. 3. The first peak appearing at the void volun~c c*ont.ained the residual glycoprotein and the second peak contained the released oligosaccharides.
The third peak, with a smaller molecular weight than the second peak, possibly consisted of t,hc degraded products of the released oligosacc:haritlcs, b~ausc its amount increased after 1)rolonged alkaline treatment. and its composition was found to 1~ almost the same as that of the second peak. The chemical coml)osition of the de~radcd #lycoprotcins and the released oligosaccharidcs are sho\vn in 'I'ablcs I and III, respectively. It was evident from the tables that substantial amounts of sialic acid, galactosc, and galactosaminc wcrc selectively removed from the ~lyc~oprotein by the alkalino tlcgradation, whereas the con- 1.0 2.5 1.5 3 0 n Aspart ic arid, scrinc, thrconinc, :md glut nmic acid were prcdominant amino acids.
Purthermorc, as sl~onii in Table I, the treatment with alkaline borol~ytlridc resulted in loss of scrine and thrconine Ivith concomitant increase in the amount OF oraminobutyric* acid. This fact is iii line with the previous obscrvation by Thomas and Winzlcr (39) that most of the alkalilabile oligosa.ccharide has the sugar sequcncc, sialic acid (2 + 3) Gal (I ---f 3) [sialic acid (2 + S)] GalSA(*, and is 0-alycosidically linked to a hydrosvl group of scrinc or threoninc.
Hemagglutinatio~ Inhibitory Activity of Alkali-treated Glycoprotein-After the alkaline, I,cductivc dcpradation, the following three types of changes in haptcnc inhibitory activity of the glycoprotein were clearly noticeable, as shown in Table II, depending on the specificity of hcmagglutinin used in the inhibition assays.
(a) The inhibitory activity against anti-11 or anti-S serum was lost already after 24 hours of the alkaline treatment in accordance with the results of Lisowka (40). Although the decrease in the inhibitory a&\-it?-against. various anti-I-1 rcagcnts during the alkalilie treatment. was less distinct than that in the case of anti-11 or anti-N serum, the glycoprotein lost its inhibitory activity against thcsc anti-11 reagents after 48 hours of the alkaline treatment, during n-hich nearly two-thirds of galactosaminc was found to be released.
-1 similar callange of inhibitory activity was also observed against 13. purpurea ancl Vi& granzinae phytohemagglutinins, both of which have been known to be specific for the fundamental structure of 1!I\ilN blood group antigens (41, 42). (b) Although the purified glycoprotcin had only weak inhibitory activity against concanavalin A, the inhibit,ory activity against this hemagglutinin was remarkably mhanccd after the alkaline treatment.
A similar increase in inhibitory activity was also observed against R. commzcnis hcmagglutinin.
Inhibition of [3H]Thymidine Incorporation in IITlnzan Peripheral Lymphocyte Xfinzulation- Table  IV summarizes the effects of the purified glycoprotcin and the products after the alkaline trcatmcnt on [311]thyn~itlinc incorporation of human peripheral lymphocytes exposed to various phytomitogcns and antilymphocyte serum. The stimulating: activity of six kinds of mitogens tested was found to be equally inhibited by the purified glycoprotein. After the alknlinc treatment, the inhibitory activity against the mitogcnic activities of PI-IA-1'I, W. jloribunda, pokcwed mitogcn, and antilymphocyte serum was not changed. In contrast, this treatment enhanced the inhibitory activity of the The purified glycoprotein and it,s alkali-treated product had no effect on the viability of the lymphocytes in the cultures without mitogen.

DISCUSSION
Isolation from human group 0 ergthrocytc membranes of a glycoprotein having various receptor activities against a number of hemagglutinins was achieved by a mild extraction method followed by successive passages through columns of DEAE-, CM-Sephadex, and Sephadex G-100. Small amounts of other glycoproteins were also observed in more retarded fractions of the DEAE-Sephadex chromatography, but they were not isolated. Among these chrpmatographies, CM-Sephadex column chromatography was particularly effective for the removal of contaminating proteins.
After a chloroform-methanol extraction in order to remove a trace of contaminating lipids, the purified glycoprotein thus obtained appeared to be a single molecular species based upon the results of disc electrophoresis, ultracentrifugation, and gel filtration in the presence of sodium dodecyl sulfate. l\lolecular weight of the purified glycoprotein was measured by sedimentation equilibrium in the presence of 1 7. sodium dodecyl sulfate and found to be 53,000. A molecular weight of 55,000 has been reported recently for the monomeric major glycoprotein from human erythrocyte membranes by Segrest et al. (43), and our preparation could be considered to be the same as their glycoprot,ein (44, 45) from its molecular weight and also from its chemical composition.
The purified glycoprotein contains 55c/ carbohydrate, of which sialic acid (23.6%), galactosamine (10.3%), and galact,ose (9.97J are the predominant sugars, with smaller amounts of glucosamine, marmose, and fucose, indicating that oligosaccharide chains of the glycoprotein are composed of a large amount of 0-glycosidically linked oligosaccharide chains, such as the one reported by Thomas and Winzler (39) and a much smaller amount of N-glycosidically linked oligosaccharide chains similar to the one isolated by Kornfeld and Kornfeld (9). Actually, after the alkaline, reductive degradation of the glycoprotein, the composition of the released oligosaccharide was in good agreement with that of the tetrasaccharide reported by Thomas and Winzler (39). Since the 0-glycosidically linked oligosaccharide chain reported by Thomas and Winaler contains 1 mole of N-acetglgalactosamine per chain, whereas the N-glycosidically linked oligosaccharide chain isolated by Kornfeld and Kornfeld contains 3 moles of N-acetylglucosamine per chain, we can estimate roughly that each 53,000 molecular weight glycoprotein contains, on the average, 26 0-glycosidically linked oligosaccharides and 4 N-glycosidically linked oligosaccharides, on the basis of the carbohydrate composition of the purified glycoprotein listed in Table I.
The purified glycoprotein has H and MN blood group activities and also carries the receptors for various phytohemagglutinins, in spite of the fact that these hemagglutinins can be classified into three groups from their specificities, as disclosed by inhibition assays using simple sugars as inhibitors (46). The multiple receptor activities of the glycoprotein indicate that the hemagglutinins bind to different parts of the same oligosaccharide chain or to different oligosaccharide chains in the same glycoprotein of the human erythrocyte membrane.
Since the purified glycoprotein was found not to be contaminated by glycolipids, within the limits of gas-liquid chromatography employed for this assay, the blood group II activity displayed by t'he purified glycoprotein further substantiated our previous proposal (10) that the ABH blood group antigens reside also in the glycoprotcin moiety of the human erythrocyte membrane and a part of the carbohydrate chain of the glycoprotein is controlled by genes relevant to ABH blood groups. Evidence for the existence of the ABH blood group antigens in glycoproteins of the human crythrocyte membrane has also been reported by other investigators (47-51).
The observation that the H and &IN blood group activities of the purified glycoprotein were diminished by the alkaline, reductive degradation clearly indicated that these blood group antigens resided in the 0-glycosidically linked oligosaccharide chains, even though the MN blood group activity was considered also to be controlled by cell membrane architecture in general (52). The fact that no remarkable decrease in the amount of fucose after the alkaline, reductive cleavage was obscrvcd, in spite of the diminished I-1 activity, could be explained by assumption that most of the fucose residues reside in the alkali-stable oligosaccharide chains and the alkali-labile oligosaccharide chains related to ABH blood groups represent relatively small proportions of the sugar chains in this purified glycoprotein.
On the other hand, it is interesting to note that the inhibitory activity of the glycoprotein against concanavalin A and R. communis hemagglutinin is greatly enhanced after the alkaline, reductive degradation.
This fact suggests that the receptor sites for these hemagglutinins reside in the alkaline-resistant oligosaccharide chains and the removal of the alkali-labile oligosaccharide chains makes these receptor sites more accessible to the hemagglutinins. The receptors for P. vulgaris, L. culinaris, V. faba, P. s&urn, S. juponica, and wheat germ hemagglutinins were also considered to reside mostly in the alkali-resistant oligosaccharide chains, possibly in N-glycosidically linked oligosaccharide chains, because the receptor activities for these hemagglutinins were not influenced by alkaline treatment of the glycoprotcin.
These results are in good agreement with our previous results (53) on the chemical nature of receptor sites for concanavalin A, L. culinaris, and P. vulguris hemagglutinins, which demonstrated that N-glycosid-ically linked oligosaccharide chains in porcine thyroglobulin esert strong inhibitory activity against these hemagglutinins. However, in view of the fact that about 407, of the N-acetylgalactosamine residues remained even after 48 hours of alkaline treatment, the possibility that a part of the receptor activity against these hemagglutinins also resides in the remaining Oglycosidically linked oligosaccharide chains cannot be completely ruled out.
The purified glycoprotein was also disclosed to be a potent inhibitor for [H]thymidine incorporation of human lymphocytes exposed to various phytomitogens and antilymphocyte serum. This finding indicates a possible similarity of oligosaccharide chains on lvmphocytc cell surface with those of erythrocyte membranes.
Furthermore, the fact that this activity was not diminished by the alkaline, reductive degration suggested that the recept,or sites for these mitogens resides in the alkali-stable Nglycosidically linked oligosaccharide chains.