The structure of a phytohemagglutinin receptor site from human erythrocytes.

Abstract A highly purified glycopeptide with potent phytohemagglutinin (PHA) receptor site activity has been isolated from human erythrocyte membranes. The glycopeptide was released from the membranes by trypsin, treated with alkaline borohydride, and purified by repeated gel filtration, further proteolytic digestion with Pronase, and diethylaminoethyl cellulose chromatography. It has a molecular weight of approximately 2000 and the following composition (in residues): sialic acid, 1; galactose, 2; mannose, 2; N-acetylglucosamine, 3; aspartic acid, 1.5; serine, 1; and threonine, 1. The sequence of the carbohydrate residues in the oligosaccharide chain was determined by sequential cleavage of the sugars from the nonreducing end with specific glycosidases. The glycopeptide has a single branched oligosaccharide chain containing two nonreducing termini, one with the composition galactose (β)/→ N-acetylglucosamine and the other with sialic acid → galactose (β)/→ N-acetylglucosamine. Each branch is connected to the inner core which contains 2 mannose and 1 N-acetylglucosamine residues. Since the oligosaccharide chain was not released from the trypsin fragment during the alkaline borohydride treatment, it is probably linked to the peptide backbone by an N-acetylglucosaminylasparagine linkage. Removal of the sialic acid residue does not affect PHA-inhibitory activity. When the galactose residues are removed, 90% of the PHA-inhibitory activity is lost. Model compounds having the terminal sequences galactose (β)/→ N-acetylglucosamine have very poor PHA-inhibitory activity unless they also have mannose residues in their interior, showing that the inner core sugars can influence PHA-inhibitory activity. The peptide backbone of the glycopeptide also affects the PHA-inhibitory activity of the oligosaccharide chain since higher molecular weight glycopeptides extracted from erythrocyte membranes with chloroform-methanol lose 90% of their PHA-inhibitory activity following proteolytic digestion with trypsin.

with potent phytohemagglutinin (PHA) receptor site activity has been isolated from human erythrocyte membranes.
The glycopeptide was released from the membranes by trypsin, treated with alkaline borohydride, and purified by repeated gel filtration, further proteolytic digestion with Pronase, and diethylaminoethyl cellulose chromatography.
It has a molecular weight of approximately 2000 and the following composition (in residues): sialic acid, 1; galactose, 2 ; mannose, 2 ; Nacetylglucosamine, 3; aspartic acid, 1.5; serine, 1; and threonine, 1. The sequence of the carbohydrate residues in the oligosaccharide chain was determined by sequential cleavage of the sugars from the nonreducing end with spectic glycosidases.
The glycopeptide has a single branched oli.gosaccharide chain contaming two nonreducing termini, one with the composition galactose -%+ N-acetylglucosamine and the other with sialic acid + galactose -% N-acetylglucosamine. Each branch is connected to the inner core which contains 2 mannose and 1 N-acetylglucosamine residues.
Since the oligosaccharide chain was not released from the trypsin fragment during the alkaline borohydride treatment, it is probably linked to the peptide backbone by an N-acetylglucosaminylasparagine linkage. Removal of the sialic acid residue does not affect PHAinhibitory activity.
When the galactose residues are removed, 90% of the PHA-inhibitory activity is lost. Model compounds having the terminal sequences galactose -4 N-acetylglucosamine have very poor PHA-inhibitory activity unless they also have mannose residues in their interior, showing that the inner core sugars can influence PHAinhibitory activity.
The peptide backbone of the glycopeptide also affects the PHA-inhibitory activity of the oligosaccharide chain since higher molecular weight glycopeptides extracted from erythrocyte membranes with chloroformmethanol lose 90% of their PHA-inhibitory activity following proteolytic digestion with trypsin.
Phytohemagglutinin extracted from the red kidney bean, Phaseolus vulgaris, has the capacity to agglutinate erythrocytes * This research was supported in part by Grants HE-00022-22 and leukocytes and to stimulate lymphocytes to undergo mitosis (1). Weber, Nordman, and G&beck (2) have separated PHAl into two glycoprotein components which stimulate mitosis in lymphocytes but differ from each other in that one component has predominantly erythroagglutinating activity while the other has predominantly leukoagglutinating activity.
We have previously demonstrated that trypsin treatment of human erythrocytes releases from the cell membrane a glycoprotein fragment which is capable of binding to the erythroagglutinating phytohemagglutinin and abolishing its erythroagglutinating and lymphocyte-stimulating properties (3). Thus, the glycopeptide has the properties expected of a cellular receptor site. The specificity for binding to erythroagglutinating PHA was shown to reside in the oligosaccharide portion of the molecule, with the major determinant sugar being a galactose residue. In this report we describe the isolation and characterization of a highly purified glycopeptide from human erythrocytes which has potent erythroagglutinating PHA-inhibitory activity. Sequential degradation with purified glycosidases has been utilized to establish the partial sequence of the carbohydrate units of the glycopeptide and to determine which carbohydrate residues are essential for binding to erythroagglutinating PHA.

EXPERIMENTAL PROCEDURE
Preparation of Phytohemagglutinin PHA from P. vulgaris was obtained as Bacto-phytohemagglutinin-P (Difco) and purified as described by Weber et al. (2). The unpurified PHA-P was ordinarily used for hemagglutination inhibition assays while the purified erythroagglutinating PHA (Peak III of Weber et al. (2)) was iodinated by the chloramine-T method (4) and used in the lymphocyte binding experiments.

Assay oj Phytohemagglutinin
Inhibitory Activity PHA receptor site activity was routinely assayed by its ability to inhibit PHA-induced hemagglutination.
One unit of inhibitory activity is defined as the amount of material necessary to inhibit completely red cell agglutination in the standard system for 3 min.

Preparation of Erythrocyte Glycopeptide
Step I: Trypsin Treatm.ent of ErythrocytesThe erythrocytes from 3 liters of outdated bank blood were washed three times with 3 volumes of 0.9% NaCI-0.01 M NaHCOz to remove the plasma and buffy coat, and then the erythrocytes were treated with tryspin as described by Winzler et al. (5). One volume of packed red cells was added to 1 volume of 0.9% NaCI-0.05 M phosphate buffer, pH 7.5, containing 0.25 mg per ml of trypsin l/250 (Difco).
The cell suspension was incubated with shaking at 37" for 1 hour during which time essentially no cell lysis oc- curred.
The cells were then removed by centrifugation and to the chilled supernatant fluid one-eighth volume of cold 50% trichloracetic acid was added. The resulting precipitate was removed by centrifugation at 15,000 X g for 10 min and the supernatant fluid, containing all the trypsin-released glycopeptides, was neutralized with NaOH, dialyzed overnight at 4", and then lyophilized.
Step 9: Alkaline Borohydride Treatment-The trypsin-released glycopeptides from 1200 ml of red cells were incubated (total volume, 15 ml) with 0.1 M sodium borohydride in 0.2 N NaOH for 36 hours at room temperature under an NS atmosphere. The solution was then neutralized with 2 N HCI and concentrated to 3 ml.
Step 8: Gel Filtration-The concentrated material from the alkaline borohydride treatment was applied to a Sephadex G-25-80 column (1.2 x 33 cm). The column was eluted with water and 1.5-ml fractions were collected.
The glycopeptide material which appeared in the exclusion volume of the column contained all the PHA inhibitory activity but only one-third of the sialic acid. This active material from three identical preparations was concentrated and applied to a Sephadex G-75-40 column (1.5 x 80 cm) (Fig. 1). Elution was carried out with water and 3.0.ml fractions were collect.ed. The active glycopeptides were combined into Fractions A and B as shown in Fig. 1.
Step 4: Pronase Digestion-The A fractions and the B fractions from four identical Sephadex G-75 columns were each combined (representing material derived from 12 preparations totaling 14,200 ml of packed erythrocytes) and concentrated to a volume of 10 ml. The two fractions (A and B) were each incubated with 15 mg of Pronase in 0.05 M Tris buffer, pH 7.8, containing 0.002 M CaClz in a final volume of 15 ml. Incubation was carried out for 66 hours at 37" under a toluene atmosphere. The reaction mixtures were then heated in a boiling water bath for 4 min and cooled. This treatment caused less than 5% loss of PHA inhibitory activity of Fraction B and approximately a 15% loss of activity of Fraction A.
Step 5: Repeat Gel Filtration-The A and B glycopeptide reaction mixtures were reduced to a volume of 3 ml and applied were performed as before. Z.U., inhibitory unit.
to a Sephadex G-50 column (1.5 x 80 cm) (Fig. 2). Elution was carried out with water and 2.5-ml fractions were collected. The A glycopeptide material contained one main peak of PHA inhibitory activity while the B glycopeptide material resolved into three active peaks (I, II, and III).
Step 6: DEAE-cellulose Chrowlatography-The B-II fraction from the Sephadex G-50 column was concentrated and loaded onto a column (0.9 x 11.5 cm) of DEAE-cellulose that had been equilibrated with 3 mM sodium phosphate buffer (pH 6.8). The column was washed with 30 ml of the equilibrating buffer and eluted with a linear gradient (160 ml) from 3 to 100 mM phosphate buffer, pH 6.8 (Fig. 3). The column was then washed with 200 mM phosphate buffer, pH 6.8. Fractions of 3.3 ml were collected.
The active fractions from each of the four peaks (II-D-l, 11-D-2, II-D-S, and 11-D-4) were pooled.
Step 7: Repeat DEAE-cellulose Chromatography-The II-D-3 fraction was concentrated, dialyzed against water, and then loaded onto another DEAE-cellulose column that had been equilibrated with 3 mM phosphate buffer, pH 6.8. The column was washed with 45 ml of the starting buffer and then successively eluted with 65 ml of 5 InM phosphate buffer, 105 ml of 7 mM phosphate buffer, and finally 35 ml of 50 InM phosphate buffer (all pH 6.8) (Fig. 4).

Chemical Determinations
Sialic acid was measured by the method of Warren (6) following hydrolysis in 0.05 N HzS04 for 1 hour at 80" (or 1 N HCl for 1 min at 100" (7)). Total hexose was measured by the phenolsulfuric acid method scaled down to one-fifth volume (8). Individual neutral sugars were determined following hydrolysis in 2 N H&O4 for 4 hours at 100" and paper chromatography in l-butanol-ethanol-water (10: 1:2) (Solvent I). Recovery was corrected for by including 1% tracer sugars in -the hydrolysis. Galactose and mannose were measured by the Park-Johnson ferricyanide method (9) and fucose by the cysteine-HsS04 method (10). Both tests were scaled down to one-fifth volume. Hexosamine was determined with a Spinco automatic amino acid analyzer following hydrolysis in 4 N HCl for 4 hours at 100" in a vacuum and lyophilization to remove HCl. N-Acetylglucosamine released by P-N-acetylglucosaminidase was measured by the method of Reissig, Strominger, and Leloir (11). Quantitative estimates of amino acid composition were obtained with the amino acid analyzer following hydrolysis of the glycopeptide in constant boiling KC1 for 16 hours at 108" in an evacuated sealed tube. were prepared from jack bean meal by a slight modification of the method of Li (12). The fi-galactosidase preparation was free of /3-N-acetylglucosaminidase and cr-mannosidase activities while the P-N-acetylglucosaminidase preparation was free of fi-galactosidase and a-mannosidase. Assays for the glycosidases were carried out in 0.05 M citrate buffer, pH 4.6; the substrates were, for P-galactosidase, 10 mM o-nitrophenyl fi-galactoside;
/?-Galactosidase assays were carried out at 37" and the others at 25". A unit of activity in each case was defined as the amount of enzyme which could liberate 1 .O pmole of o-or p-nitrophenol in 60 min.

Action of Glycosidases on Glycopeptides
Because of the limited amount of purified glycopeptide that was available, the following scheme was used during the sequential enzyme treatment of the purified glycopeptide.
The reaction mixture (0.2 ml) was incubated at 37" under a toluene atmosphere for a specified number of hours and then heated in a boiling water bath for 2 min. Water and tracer 14C sugars (galactose and mannose) were added to the heated reaction mixture to bring the volume to 0.5 ml and the precipitated enzyme protein was removed by centrifugation.
The supernatant fluid was then applied t,o a previously calibrated Sephadex G-25 fine column (0.9 x 9.0 cm) (V,, = 2.5 ml) to separate the residual glycopeptide (in the exclusion volume) from the released sugar residues (which were retarded).
The residual glycopeptide, which was recovered quantitatively, was evaporated to dryness, redissolved in 0.05 M citrate buffer, pH 4.6, and reincubated with the same glycosidase.
This process was repeated until no more sugar was released (two incubations were usually adequate) and then the residual glycopeptide was incubated with another glycosidase.
When release of galactose or mannose was being measured the released sugar residues plus the tracer 14C sugars were deionized with an Amberlite MB-3 mixed bed resin, evaporated to dryness, and then spotted on Whatman NO. 1 paper and chromatographed in Solvent I. The released sugars were quantitatively determined as described above. When release of N-acetylglucosamine was being measured an aliquot of the retarded fraction was analyzed directly.
In this manner serial incubations could be performed on a single glycopeptide sample.
In some experiments N-acetylglucosamine release was followed

Chloroform-Methanol Extraction of Erythrocyte Ghosts
Washed red cells were lysed with 9 volumes of water and sedimented at 12,000 x g for 12 min. The packed ghosts were then washed four times with 0.01 M Tris-0.0001 M EDTA, pH 7.4, to remove residual hemoglobin.
For every milliliter of packed "ghosts," 9 ml of chloroform-methanol (2 : 1) were added with vigorous shaking of the flask. The mixture was stirred for 30 min and the aqueous layer was carefully removed and concentrated in a rotary evaporator.
Glycopeptides and Oligosaccharides-The transferrin glycopeptide containing a single asparagine residue was a gift of Dr. Graham Jamieson (13). Lacto-N-tetraose, lacto-N-neotetraose, and lactosamine were kindly provided by Drs. Kobata and Ginsburg (National Institutes of Health). Fetuin was purchased from Grand Island Biological Company (Grand Island, New York) and the fetuin glycopeptides isolated by Sephadex gel filtration following Pronase digestion as described by Spiro (14).

Isolation of Glycopeptides with Phytohemagglutinin
Inhibitory Activity-Tryspin treatment of intact human erythrocytes released from the cell membrane a number of glycopeptides which were capable of binding to erythroagglutinating PHA and inhibiting the agglutination of red cells by that molecule. When the trypsin-released glycopeptides were treated with alkaline borohydride, approximately two-thirds of the sialic acidcontaining oligosaccharides were released from the peptide backbone (presumably those chains linked 0-glycosidically to the hydroxyl groups of serine and threonine (15)). The residual glycopeptide material, which eluted in the exclusion volume of the Sephadex G-25 column, retained almost full inhibitory activity.
This material was separated into two main corn ponents (A and B) upon subsequent gel filtration on Sephadex G-75 (Fig. 1). When the active glycopeptides in Peaks il and B were subjected to proteolytic digestion with Pronase followed by Sephadex G-50 gel filtration, the patterns seen in Fig. 2 were obtained.
The A glycopeptides resolved into several sialic acid-containing peaks, only one of which had PHA inhibitory activity.
However, chromatography of this active material on DEAE-cellulose revealed that it contained at least five active components.
Since the amount of glycopeptide material in these fractions was quite limited (compare Peaks A and B of Fig. I), further purification steps were not undertaken.
The Pronase-treated B glycopeptides separated into three active fractions (I, II, and III) on Sephadex G-50 (Fig. 2). When each of these fractions was subjected to DEAE-cellulose chromatography, each was resolved into a number of active components.
Since the most highly purified glycopeptide with PHA inhibitory activity was derived from the Peak II material, the subsequent description of the purification scheme will be limited to this material.
Chromatography of the B-II material on DEAE-cellulose resulted in the separation of a number of sialic acid-containing glycopeptides as shown in Fig. 3. Only two of these glycopeptides had significant PHA inhibitory activity while others, such as the material eluted with the 0.2 M phosphate buffer, were almost completely inactive.
The active glycopeptides which eluted from the column when the conductivity of the phosphate buffer was 0.55 millimho were combined (designated 11-D-3), dialyzed, and rechromatographed on DEAE-cellulose (Fig. 4). Batch elution with 5 mM phosphate buffer, pH 6.8 (conduct'ivity of 0.55 millimho), resulted in the separation of this glycopeptide from another active, but more negatively charged glycopeptide.
The glycopeptide (11-D-3) which eluted from the DEAEcellulose column with the 5 mM phosphate buffer contained a total of 27,500 PHA inhibitory units, representing a recovery of about 3.7% of the total inhibitory activity in the starting material.
The data in Table I summarize the recovery of inhibitory activity during the various stages of purification of 11-D-3.
Composition of Glycopeptide II-D-S-The composition of the most highly purified glycopeptide is shown in Table II. The molecular weight of the glycopeptide would be approximately 2000 based on the chemical composition and assuming a single oligosaccharide chain. When II-D-3 was subjected to gel filtration on Sephadex G-50, it was more retarded than a purified transferrin glycopeptide with a molecular weight of approximately 2500 (13). This finding is consistent with the proposal that II-D-3 has a single oligosaccharide chain. The fact that the oligosaccharide chain of the glycopeptide had not been released during the alkaline borohydride treatment of the original trypsin fragment suggests that it is linked to the peptide through a glycosylamine-type linkage between asparagine and N-acetylglucosamine.
Action of Glycosidases on II-D-S-When the II-D-3 glycopeptide was treated successively with fl-galactosidase and P-Nacetylglucosaminidase, 0.9 residue of galactose and 0.9 residue of N-acetylglucosamine were released per oligosaccharide chain (Table III).
Treatment of the glycopeptide with P-N-acetylglucosaminidase alone released only 0.12 residue of N-acetylglucosamine.
These results indicated that the oligosaccharide chain of II-D-3 contains a nonreducing end with the sequence galactose -% N-acetylglucosamine -% X. When the terminal sialic acid residue of II-D-3 was removed by mild acid hydrolysis (desialized II-D-3 in Table III) and then the sequential enzyme treatment was carried out, 1.7 residues of galactose and 2.0 residues of N-acetylglucosamine were released. Since 1 galactose residue and 1 N-acetylglucosamine residue are contained in the nonreducing end just described, the remaining galactose and N-acetylglucosamine residues must be in a second chain terminating in sialic acid which has the sequence sialic acid + galactose 4 N-acetylglucosamine -% Y. While the galactose residues were released relatively rapidly (9 hours incubation time), the N-acetylglucosamine residues were released at a much slower rate (Fig. 5). However, even after prolonged digestion, no more than 0.9 and 2.0 residues of N-acetylglucosamine were released from II-D-3 and desialized 11-D-3, respectively.
The release of galactose from II-D-3 and desialized II-D-3 by /3-galactosidase was accompanied by a loss of PHA inhibitory activity.
The amount of inhibitory activity lost was proportional to the amount of galactose liberated at intermediate time intervals and after completion (9 hours) II-D-3 retained 327, of its original activity and desialized II-D-3 only 13% of its original activity.
These results showed the importance of the galactose residues in the binding of the glycopeptide to PHA. Removal of the sialic acid residue did not result in a loss of PHA inhibitory activity.
The glycopeptide did not release any mannose when treated         initially with a-mannosidase. When II-D-3 was treated successively with &galactosidase, fl-A-acetylglucosaminidase, and then with cY-mannosidase, 0.32 residue of mannose was released (Table III).
Similar treatment of desialized II-D-3 resulted in the release of 1.7 residues of mannose.
These results show that (a) the mannose residues are in the inner core of the oligosaccharide chain, (b) the terminal chain with the sequence sialic acid + galactose + N-acetylglucosamine protects most but not all of the mannose residues, and (c) the mannose residues can be released from the inner core with or-mannosidase when both of Periodate Oxidation of II-D-S-Periodate oxidation of the glycopeptide resulted in marked destruction of the galactose and mannose residues but no significant destruction of N-acetylglucosamine (Table IV).
Binding of Glycopeptide II-D-S to Phytohemagglutinin-The glycopeptide II-D-3 binds specifically to phytohemagglutinin as shown in Fig. 6. In this experiment phytohemagglutinin, but not albumin, was able to displace the glycopeptide from its usual elution position on Sephadex G-75 to the exclusion volume of the column.
The data also indicate that at least 71 y0 of the glycopeptide material is capable of binding to phytohemagglutinin, thus providing an estimate of the purity of the glycopeptide preparation.
Previous incubation of the glycopeptide with PHA for a longer time (3 or 5 hours) did not increase the amount of II-D-3 bound.
When the unbound material was reisolated and incubated again with PHA only a few percentage of it bound, indicating that II-D-3 contains a small amount of glycopeptide material that is incapable of binding to PHA.
This inactive material contains sialic acid, galactose, and mannose in amounts roughly proportional to those found in the over-all composition of II-D-3 (Table II).
However, the amount of inactive material recovered was so small that precise quantitative determination could not be performed.
In all, six different binding studies were performed with different concentrations of PHA. The amount of glycopeptide bound to PHA was measured and the amount of free PHA and free glycopeptide were calculated from the known amounts of each initially present.
The association constant, K = (glycopeptide bound to PHA)/(free PHA) (free glycopeptide), was then calculated and the average value was 4.7 X 103~-1. Effect of Glycopeptide II-D-S on Phytohemagglutinin Binding to Lymphocytes-The standard hemagglutination inhibition assay for PHA inhibitory activity tests only the ability of the glycopeptide to inhibit PHA binding to erythrocytes. Previously we have shown that a partially purified glycopeptide preparation from erythrocytes is also capable of inhibiting PHA binding to lymphocytes and of abolishing the mitogenic response induced by PHA (3). As shown in Fig. 7, the glycopeptide II-D-3 is a potent inhibitor of the binding of purified i311-erythroagglutinating PHA to lymphocytes. Using the data in Fig. 7 and knowing the amount of l%PHA bound by these same lymphocytes at various concentrations of PHA it was possible to calculate an approximate association constant of PHA for II-D-3 (K = 4.5 X lo3 M-l) which agreed very well with the value from the direct binding studies.
Comparison of Phytohemagglutinin Inhibitory Activity of various Oligosaccharides-The sugar sequence studies (Table  III) show that the glycopeptide II-D-3 has an oligosaccharide chain containing the sequence sialic acid ---t galactose --f Nacetylglucosamine with the galactose residue being essential for PHA inhibitory activity.
As this is a fairly common sugar sequence in various oligosaccharides, we examined the ability 2542 Phytohemagglutinin Receptor Site   Table V. The glycopeptide II-D-3 had action of the trypsin in splitting the glycopeptide from the the most potent PHA inhibitory activity of the compounds erythrocyte cell surface was producing a fragment with reduced tested. Both the fetuin and transferrin glycopeptides had PHA binding activity compared to the intact site on the cell considerable inhibitory activity.
On the other hand, the oligo-surface, another technique was employed to solubiliae the PHA saccharides lacto-N-tetraose, lacto-N-neotetraose, and lactosa-receptor site in a less "degraded" state. When erythrocyte mine, all of which contain the sequence galactose + N-ghosts were extracted with chloroform-methanol, virtually all acetylglucosamine, had less than 0.1% the inhibitory activity of the sialic acid-containing material was solubilized into the of 11-D-3.
These data suggest that the inner sugars of the aqueous layer in the form of a large molecular weight glycoprooligosaccharide chain, particularly the mannose residues, are also tein (Fig. 8). This material, which was totally excluded from involved in the binding of PHA to the oligosaccharide.
The Sephadex G-75 (mol wt > 50,000), had high levels of PHA inhibisimple sugar N-acetylgalactosamine had a slight inhibitory tory activity (64,000 inhibitory units derived from 100 ml of activity at very high concentrations but N-acetylglucosamine erythrocyte ghosts). Fig. 8 also shows that the usual trypsin proand galactose were virtually devoid of activity in this assay cedure when applied to 100 ml of the erythrocyte ghosts liberated system. only 30% of the sialic acid in the form of lower molecular weight  glycopeptides with only 3,500 inhibitory units. When the trypsin-treated ghosts were then extracted with chloroform-methanol, the aqueous layer contained all the remaining sialic acid in high molecular weight glycopeptides having only 5,375 inhibitory units.
Thus, the trypsin treatment had left behind on the ghosts the majority of their sialic acid but less than 10% of their original chloroform-methanol extractable PHA inhibitory activity.
The most logical explanation for the fact that the total activity in the two fractions (trypsin-released plus chloroformmethanol-extracted) was so much less than the total activity in the direct chloroform-methanol extraction of ghosts, was that proteolytic digestion by trypsin could reduce the activity of the receptor site. To test this possibility, the large molecular weight glycoproteins from the direct chloroform-methanol extraction were treated with trypsin and found to be degraded to lower molecular weight glycopeptides (retarded on Sephadex G-75) which retained only 10% of the PHA inhibitory activity or 6,400 inhibitory units. Therefore, it can be concluded that the glycopeptides released from the red cell surface by trypsin do represent fragments of a larger glycoprotein moiety of the surface membrane which when intact has a much greater affinity for binding to PHA.
The final product by both routes (trypsin or chloroform-methanol + trypsin) is a small molecular weight glycopeptide of similar activity but the trypsin method releases fragments more selectively so that the active glycopeptides are much less contaminated with inactive sialic acid-containing glycopeptides. DISCUSSION We have previously demonstrated that trypsin treatment of human erythrocytes releases from the cell membrane glycopeptide fragments which retain their capacity to bind to PHA (3). These glycopeptides have multiple oligosaccharide chains, of which only a portion are capable of binding to PHA (3). By using a combination of alkaline borohydride treatment, Pronase digestion, repeated gel filtration, and DEAE-cellulose column chromatography we have been able to isolate a purified glycopeptide which has potent PHA inhibitory activity. This glycopeptide fraction contains about 25 y0 inactive molecules (i.e. cannot bind to PHA) which have the same gross composition as the active molecules.
They must also have a structure very similar to that of the active molecules since the degradation studies on the glycopeptide fraction II-D-3 with fl-galactosidase and fi-N-acetylglucosaminidase gave results expected of an essentially homogeneous material.
On the basis of the data presented in this paper it can be concluded that glycopeptide II-D-3 has a single, branched oligosaccharide chain. The experimental evidence supports the following structure of the oligosaccharide chain. 1. One of the branches terminates in a galactose residue which can be removed with P-galactosidase to expose an N-acetylglucosamine residue which in turn can be removed by fi-N-acetylglucosaminidase.

Galactose
-% N-acetylglucosamine 4 2. The other branch terminates in sialic acid. Removal of the sialic acid exposes an additional galactose residue which can be removed by P-galactosidase, to expose an additional Nacetylglucosamine residue susceptible to cleavage by P-Nacetylglucosaminidase.
Sialic acid -% galactose -$ N-acetylglucosamine 4 3. The inner core of the oligosaccharide after removal of the two termini (1 sialic acid, 2 galactose, 2 N-acetylglucosamine residues) contains 2 mannose residues and 1 N-acetylglucosamine. The fact that neither of the mannose residues had been released by alkaline borohydride treatment suggested that they are not linked 0-glycosidically to the serine or threonine residues of the peptide backbone.
Further, since both of the mannose residues previously observed that there is a high content of aspartic acid, serine, and threonine in the glycopeptides released by trypsin from the erythrocyte membrane.
The ability of the glycopeptide II-D-3 to act as an inhibitor of PHA is determined by a number of factors.
When the galactose residues are removed by /%galactosidase, the compound loses 90% of its inhibitory activity, indicating the importance of these sugar residues. The fact that removal of a galactose residue from only one chain results in a greater than 50% loss of PHA inhibitory activity suggests either that the activity of the two branches is not equal or that the two branches together are more active than either is alone. Whether or not the galactose residue is covered by a sialic acid residue is relatively unimportant as far as PHA inhibitory activity is concerned. We have previously reported the occurrence of a glycopeptide isolated from Peak A on Sephadex G-75 in which all the galactose residues were penultimate to sialic acid residues (3). The PHA inhibitory activity of that glycopeptide was not affected by the removal of the sialic acid residues. The studies utilizing model compounds (Table V) suggest that the inner sugars of the oligosaccharide chain, most likely the mannose residues, also affect the PHA inhibitory activity of the glycopeptide. Thus, of those compounds with terminal galactose --) N-acetylglucosamine sequences, only those with mannose in the inner portion of the oligosaccharide chain had any significant biologic activity. The mannose residues could be important in at least two ways. First, they could actually be a part of the receptor site which PHA recognizes and binds to. This would be similar to the Leb blood group determinant which consists of 2 fucose residues (23). Secondly, the mannose residues could affect the spatial arrangement of the galactose residues, presumably holding them in a more favorable conformation. Molecular models of II-D-3 show that the mannose residues could act in both of the abovementioned ways.
The biological activity of the oligosaccharide chain is also affected by the nature of the protein backbone.
As shown in Fig. 8, trypsin digestion of a high molecular weight glycopeptide results in a 90% loss of PHA inhibitory activity. The most likely explanation for this is that the protein backbone holds the oligosaccharide chains in a favorable conformation. An analogous situation is found in the case of soluble blood group substances.
In these high molecular weight glycoproteins the by guest on March 22, 2020 http://www.jbc.org/ Downloaded from specific determinants are located in the carbohydrate chains but the protein backbone has a major effect on the reactivity of the molecule.
Thus, if these molecules are subjected to proteolytic digestion, a marked loss of biologic activity occurs (24).
During the course of purification of II-D-3 numerous other glycopeptide fractions with PHA inhibitory activity were also separated.
In order to determine whether the structures of these active glycopeptides were similar to 11-D-3, a few such fractions have been further purified and subjected to structural analysis2 The carbohydrate portion of each active glycopeptide examined has in common with glycopeptide II-D-3 a branched oligosaccharide chain with a structure similar to that shown in Fig. 9. In addition some of these glycopeptides contain significant amounts of fucose and an over-all composition consistent with the presence of an additional carbohydrate chain of different structure.
Further, these glycopeptides differ in the number and kind of amino acid residues left on the glycopeptide fragments following the Pronase digestion step. These features, combined with the fact that the branched oligosaccharide moiety can contain either 1 or 2 terminal sialic acid residues, most likely account for the size and charge heterogeneity displayed by active glycopeptides.
The finding that the highly purified glycopeptide II-D-3 retains the ability to inhibit PHA binding to lymphocytes suggests that the erythroagglutinating PHA may bind to a similar oligosaccharide on both the lymphocyte and erythrocyte cell surf aces. We are currently attempting to establish this point directly by isolating the PHA receptor site from lymphocyte membranes.