Topography and Polypeptide Distribution of Terminal N-Acetylglucosamine Residues on the Surfaces of Intact Lymphocytes EVIDENCE FOR O-LINKED GlcNAc*

Bovine milk galactosyltransferase has been used, in conjunction with UDP-[3H]galactose, as an imper- meant probe for accessible GlcNAc residues on the surfaces of lymphocytes. Galactosylation of living thymic lymphocytes is depedent upon cell number, en- zyme concentration, UDP-galactose concentration, and Mn2+ concentration. Kinetics of labeling are bi- phasic, leveling off at approximately 30 min. The data strongly indicate vectorial surface labeling and cova- lent attachment of galactose. Thymocytes, T-lympho-cytes, and B-lymphocytes have approximately lo‘, 3 % lo‘, and 5 X 10’ galactosylatable sites on their cell surfaces, respectively. Numerous proteins are exoga- lactosylated that differ quantitatively among the major functional subsets of lymphocytes. Negligible radioac- tivity is found in lipid. alkali labile, whereas T-lympho-cytes which

Bovine milk galactosyltransferase has been used, in conjunction with UDP-[3H]galactose, as an impermeant probe for accessible GlcNAc residues on the surfaces of lymphocytes. Galactosylation of living thymic lymphocytes is depedent upon cell number, enzyme concentration, UDP-galactose concentration, and Mn2+ concentration. Kinetics of labeling are biphasic, leveling off at approximately 30 min. The data strongly indicate vectorial surface labeling and covalent attachment of galactose. Thymocytes, T-lymphocytes, and B-lymphocytes have approximately lo', 3 % lo', and 5 X 10' galactosylatable sites on their cell surfaces, respectively. Numerous proteins are exogalactosylated that differ quantitatively among the major functional subsets of lymphocytes. Negligible radioactivity is found in lipid.
In thymocytes, 49% of the exogalactosylated oligosaccharides are alkali labile, whereas 80 and 90% of that derived from T-lymphocytes and B-lymphocytes can be @-eliminated, respectively. Sensitivity of the intact proteins or tryptic peptides to the peptide:N-glycosidase also confirms the relative amounts of cell surface, N-linked and O-linked oligosaccharides which are exogalactosylated. Composition, size, and high performance liquid chromatography on two types of high resolution columns establish that the bulk of the exogalactosylated, @-eliminated oligosaccharides are Gal@l-4GlcNAcitol. These data suggest the presence of O-glycosidically linked GlcNAc monosaccharide on many lymphocyte cell-surface proteins. However, additional experiments indicate that the majority of these moieties appear to be cryptic or inside the cell. Thus, these studies not only describe dramatic differences in the amounts and distribution of terminal GlcNAc residues on phenotypically different lymphocyte populations, but they also describe the presence of a novel protein-saccharide linkage, which is present on numerous lymphocyte proteins.

~~
Nearly all of the characterized lymphocyte differentiation or histocompatability antigens are cell-surface glycoproteins (1). In fact, the expression of specific membrane glycoproteins has proven to be the most reliable marker for functionally distinct subsets of lymphocytes. Functional subsets of lymphocytes can readily be separated using monoclonal or polyclonal antibodies to these specific membrane glycoproteins (2). Interestingly, similar functional subsets of lymphocytes can also be isolated solely based upon their specific reactivity to glycosyl-binding plant lectins (3)(4)(5).
The biological functions of lymphocytes are mediated by a complex network of cellular interactions in which these cellsurface glycoproteins are directly involved (6,7). Recent work suggests that the saccharide moieties on specific lymphocyte glycoproteins play an essential role in their biological functions (8,9). Several studies have demonstrated pronounced changes in the saccharide topography or in the glycosylation of particular proteins on the surfaces of lymphocytes accompanying lymphocyte activation or development (10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20)(21). Most of these earlier studies have involved the use of lectin binding (10,15), galactose oxidase/sodium borotritide surface labeling (13, 21), whole cell electrophoresis (16, 20), or gel electrophoresis (17,19). An understanding of the changes in oligosaccharide structural diversity accompanying lymphocyte activation and development represents a first step toward ascertaining the details of the structural-functional significance of the varied glycosylation of these specific lymphocyte cellsurface receptors.
Only recently has it become feasible to use purified glycosyltransferases as specific vectorial modifiers of saccharide structure and as probes for exposed saccharide moieties on intact cells (22). This is largely due to the development of techniques, based upon affinity chromatography, for the purification of many of these enzymes (22)(23)(24)(25)(26)(27). Glycosyltransferases are remarkably specific for both their acceptor substrates and their donor sugar nucleotide. In fact, there may be a specific glycosyltransferase for the synthesis of every type of glycosidic linkage found in nature (28). Recent work has directly demonstrated the practical use of purified glycosyltransferases as vectorial cell-surface probes and as modifiers of oligosaccharide structures (22,(29)(30)(31).
Bovine milk galactosyltransferase (EC 2.4.1.38), in the absence of a-lactalbumin (32), catalyzes the following reaction, where R can be any of several different moieties.
UDP-galactose + GlcNAc-R + Ga@1-4GlcNAc-R + UDP initial studies, we have demonstrated pronounced differences in the disposition of terminal GlcNAc residues on the surfaces of lymphocyte subpopulations having functionally different phenotypes. Additionally, the data strongly indicate that the bulk of these terminal GlcNAc residues, which are exogalactosylated, originally existed as single monosaccharide units 0-glycosidically linked to protein.

RESULTS
Conditions for the Exogalactosylation of Living Lymphocytes-The conditions for the use of highly purified bovine milk galactosyltransferase as an impermeant probe of lymphocyte cell-surface saccharide topography were established. Fig. 1 summarizes the results of experiments which determined the optimal conditions for galactosylation of living lymphocytes (90-99% viable by Trypan blue exclusion). The labeling of lymphocyte surface proteins was shown to be dependent upon UDP-galactose, cell concentration, and MnZ+ ion concentration. Likewise, the reaction was dependent upon galactosyltransferase concentration, and the kinetics of labeling were biphasic, leveling off a t about 30 min. These biphasic kinetics, as shown in Fig. ID, were very reproducible from three totally independent experiments, suggesting that about 60% of the galactosylatable proteins of thymocytes were labeled more rapidly than the remaining 40% of the surface acceptors. However, in these initial studies, we chose conditions for saturation labeling of galactosylatable surface components in order to assess the topography of terminal GlcNAc residues on the surfaces of lymphocytes which have different functional phenotypes.
Prior autogalactosylation of the transferase (47), and the presence or absence of heat-inactivated serum in the labeling buffer had no effect on either the amount of incorporation, the SDS-PAGE profiles, or pronase glycopeptide profiles of the labeled lymphocyte components (data not shown). In addition, the levels of [3H]galactose incorporated into galactosyltransferase itself were insignificant compared to that incorporated into lymphocyte glycoproteins. Identical levels of incorporated radioactivity and SDS-PAGE or pronase glycopeptide profiles were obtained with both a-lactalbumin affinity purified galactosyltransferase and with the commercially available enzyme (data not shown).
Evidence for Vectorial Surface Labeling-Results of several experiments strongly indicate that virtually all of the radioactivity incorporated into lymphocytes is the result of vectorial galactosylation at the cell surface. 1) Addition of a 1000fold excess of unlabeled galactose (10 mM) to the standard labeling buffer had no effect on the incorporation of radioac-' Portions of this paper (including "Experimental Procedures" and tivity from UDP-[3H]galactose catalyzed by galactosyltransferase (control, 6.3 X lo5 cpm/106 cells; experimental, 5.7 X io5 cpm/1O6 cells; see Fig. 1). 2) Cells incubated with 10 pCi of [3H]galactose (same specific activity as sugar nucleotide normally used) incorporated insignificant amounts of radioactivity (less than 4% of that incorporated by exogalactosylation), which was totally abolished by the presence of unlabeled galactose, as in 1) above. 3) In vectorially labeled cells, virtually 100% of the incorporated radioactivity was still in the form of galactose, indicating that no epimerization had occurred. 4) Examination of total cellular associated radioactivity after standard labeling conditions indicated that generally greater than 96% was acid insoluble when cells were vectorially labeled with UDP-[3H]galactose and galactosyltransferase, whereas less than 5% of the cell-associated radioactivity was acid insoluble when an equivalent amount of [3H] galactose was used. 5) Addition of 0.1 mM galactose l-phosphate or 2 mM UDP-GlcNAc had little or no effect on incorporation of radioactivity by exogalactosylation (control, 6.3 X lo5 cpm/106 cells; galactose 1-phosphate, 6.3 x lo5 cpm/106 cells; UDP-GlcNAc, 4.5 X IO5 cpm/106 cells). 6) Under standard incubation conditions, greater than 80% of the added UDP-[3H]galactose remained intact and inhibition of its hydrolysis by 20% by the addition of 5'-AMP (2.5 mM) had no effect on levels of radioactivity incorporated. 7) Cell viability was typically greater than 95% after the galactosylation reaction. 8) Autoradiography of labeled lymphocytes (greater than 95% viable) indicated that at least 35% of the cells were heavily labeled. This percentage of labeled cells is a minimal estimate, since in the technique employed only the coverslip contained photographic emulsion (48). All of the radioactivity observed in these autoradiographic analyses of exogalactosylated lymphocytes was localized in massive "caps" at the poles of the cells rather than being evenly distributed along the periphery (manuscript in preparation). These data clearly indicate that our labeling results are not due to the selective galactosylation of small numbers of lysed or dead cells. However, it is likely that exogalactosylation is preferentially labeling certain subclasses or developmental stages of lymphocytes (see below).
Vectorial Galactosylation by Endogenous Versus Exogenously Added Galactosyltransferase-The levels of galactose incorporated from UDP-[3H]galacto~e into membrane proteins of thymocytes were at least 10-20 times higher in the presence of exogenously added galactosyltransferase (Fig. 1C).
Nature of the Membrane Associations of Vectorially Galactosylated Proteins-Exogalactosylated thymocytes were subjected to different extraction procedures to evaluate whether the galactosylated surface proteins were integral or more peripheral membrane components (50).
Secreted or very loosely associated proteins, which could be washed off by isotonic labeling media, were not examined in these studies. Nearly all of the cell-associated radioactivity was on proteins, since ch1oroform:methanol (2:l) extraction removed only 3%. Only 4% of the cell-associated, galactosylated products were extractable by either hypotonic buffer or by 5 mM EDTA (under conditions of cell lysis). When labeled thymocytes were extracted with NP-40, 34% of the radioactivity was initially solubilized with 0.05% detergent and consecutive extraction with 0.1% NP-40 increased the solubilized radioactivity by only 9%. Virtually all of the radioactivity was solubilized by SDS-PAGE sample buffer. Taken together, these data indicate that the bulk of the cell-associated lymphocyte surface proteins, which are externally galactosylatable, are integral membrane proteins.
Differential Exogalactosylation of B-, T-, and Thymic Lym-phocytes-Purified B-, T-, and thymic lymphocytes were isolated and assessed for purity by fluorescent antibodies. Thymocyte preparations were routinely greater than 99% lymphocytes by microscopic examination (33), T-lymphocytes were typically 95-98% pure, and B-lymphocytes were generally 95-100% pure. The viability of these purified cell populations, in these particular experiments, ranged from 98-99, 95-98, and 53-90% for thymocytes, T-lymphocytes, and Blymphocytes, respectively. The enzyme dependence for the exogalactosylation of purified populations of these lymphocyte subsets is shown i n Fig. 2. Nearly identical results are obtained whether the data is normalized to protein or to cell number. Data from six different experiments indicate that Tlymphocytes show an average of at least 3 X (range, 2.8-3.9X) more galactosylation than do thymocytes from the same mice. Similarly, data from five independent experiments show that B-lymphocytes have an average of a t least 5~ (range, 4.9-5.8X) more galactosylatable cell-surface sites than thymocytes.
The SDS-PAGE analysis of 3H-galactosylated proteins ( Fig. 3) indicates that numerous proteins are labeled in all three cell types. In each case, the major labeled protein has a molecular weight of 76-78,000. In addition to numerous minor components, thymocytes contain three major protein bands which are labeled by external galactosylation of living cells ( M , = 125,000, 105,000, and 76,000, respectively). All of these bands are also detectable in T-and B-lymphocytes. T-lymphocytes show four species in the 140K-105K range, and six bands at 78K, 69K, 50K, 26K, 24K, and 19K. Three of these bands (69K, 50K, and 19K) are not detectable in thymocytes, even with prolonged autoradiographic exposures. B-lymphocytes contain more labeled proteins of higher molecular weight, having six such bands, two of which (122K and 84K) are not detectable in either T-lymphocytes or in thymocytes from the same mice. Major quantitative differences among thymocytes and the splenic-derived lymphocytes are also seen in the proteins banding in the 30,000-50,000 molecular weight range. Band by band comparisons of these gels also shows numerous other quantitative differences in the labeling of specific size classes of proteins between the three types of lymphocytes (Fig. 3). The complexity of these labeling patterns and those obtained with lactoperoxidase cell-surface iodination or metabolic labeling with [35S]methionine has thus far precluded our determining whether these lymphocyte class-specific galactosylation patterns are the result of differences at the polypeptide level, glycosylation differences, or a combination of both. In any case, there appears to be profound quantitative and perhaps even some qualitative differences in the numbers and locations of cell-surface, terminal GlcNAc residues on functionally different subsets of lymphocytes.
Dfmonstration That Vectorially Added [3H]Galactose Is Covalently Attached to Protein-Several experiments were undertaken to ensure that the radioactivity was, in fact, covalently attached to protein. Evidence that the disaccharide results from galactosylation of a covalently attached moiety is summarized below. 1) Virtually all of the radioactivity on intact, detergent-solubilized proteins migrates in the void volume of Sephadex G-50, even in the presence of 4 M guanidine HC1.2) P-Elimination of material excluded from Sephadex G-50 in the presence of 4 M guanidine HC1, followed by Bio-Gel P-4 chromatography of the reaction mixtures, yields nearly identical profiles as that shown in Fig. 6. 3) DPCCtrypsin treatment of total reduced and alkylated, 3H-galactosylated proteins yielded a very complex profile of tryptic peptides on reverse phase HPLC (C-l8), which still contained nearly all of the initial protein-associated [3H]galactose, especially when normalized for typical recoveries from these HPLC columns (41,53). 4) Nearly all of the protein-associated [3H]galactose was recovered as limit pronase glycopeptides when exogalactosylated lymphocytes were exhaustively digested with pronase (Fig. 4). 5) Virtually all of the [3H] galactose on exogalactosylated lymphocytes was acid insoluble and migrated as typical protein bands upon SDS-PAGE.
Nature of the Galactosylated Oligosaccharides on Lymphocyte Cell Surfaces-In initial experiments, purified lymphocytes were exogalactosylated, and the acid-soluble cellular components were subjected to exhaustive digestion with predigested pronase (40) in order to study the size classes of labeled glycopeptides. As shown in Fig. 4, A and C, high resolution Bio-Gel P-4 (-400 mesh) chromatography indicates that most of the incorporated radioactivity in B-lymphocytes is found in four to six major limit pronase glycopeptide size classes, all of which are relatively low in molecular weight. On the other hand, thymocytes and T-lymphocytes yielded a higher proportion of larger molecular weight pronase glycopeptides (Fig. 4B). Repeated pronase digestion or even exhaustive sequential digestion of these glycopeptides with Aminopeptidase M (51) and carboxypeptidase Y (52) failed to substantially alter the gel-filtration patterns of the lower molecular weight glycopeptides (Fig. 40). In spite of their resistance to proteolysis, these glycopeptides, in fact, contain considerable peptide heterogeneity when examined by high resolution reverse phase HPLC under conditions relatively insensitive to the nature or size of the attached oligosaccharide (41, 53). It seems likely that the resistance to complete proteolysis is conferred by glycosylation.
Labeled proteins from all three cell types were analyzed for the nature of the protein-saccharide linkage containing the [3H]galacto~e by treatment with dilute alkali in the presence of excess NaBH4 (44). Fig. 5 (miniprint) shows the time course of this alkali catalyzed p-elimination of the 3H-galactosylated, intact proteins of splenic lymphocytes (mixture of T-and Blymphocytes). Surprisingly, a large proportion of the radioactivity very rapidly becomes p-eliminated under these con-  ditions. Fig. 6 shows the results of Bio-Gel P-4 (-400 mesh) chromatography of @-eliminated, 3H-galactosylated proteins of thymocytes, T-lymphocytes, and B-lymphocytes. B-lym-phocytes, T-lymphocytes, and thymocytes contained less than lo%, less than 20%, and approximately 49% of the labeled ['Hlgalactose, respectively, in oligosaccharides which were resistant to @-elimination. In fact, the bulk of the radioactivity from all three cell types was @-eliminated and exactly coeluted with authentic Gal@l-4GlcNAcitol on these high resolution gel-filtration columns.
Kinetic studies, using Bio-Gel P-4 (-400 mesh) to assay @elimination of pronase glycopeptides from exogalactosylated splenic lymphocytes (Fig. 7), indicated that no intermediate oligosaccharides, larger than the disaccharide, were detectable, even at the earliest times of the reaction when only a portion of the Gnlgl-4GlcNAcitol was released (Fig. 7, B-D).
Although similar kinetic studies on intact exogalactosylated proteins indicated that they were @-eliminated even more rapidly, again only the disaccharide was released (Fig. 5, miniprint). The total absence of side products or intermediates in both of these kinetic experiments strongly argues against "peeling" or nonspecific endohydrolytic cleavage of glycosidic bonds. This is especially true considering the very different rates of reaction for the @-elimination of intact proteins uersus pronase glycopeptides (see Figs. 5 and 7). In contrast, @-elimination of exogalactosylated asialoagalactofetuin, or asialoagalactotransferrin (which werre heavily radiolabeled) did not result in the release of a significant amount of radioactivity the size of the observed disaccharide (data not shown). Besides adding additional support to the specificity of the @-elimination reaction, the studies with asialoagalactofetuin also indicate that the bovine milk enzyme is not galactosylating terminal GalNAc residues, which is consistent with its known substrate specificity. In preliminary studies, we have also shown that the enzyme used here has no detectable activity against free GalNAc (data not shown).
When intact proteins from these 'H-galactosylated lymphocytes were treated with the enzyme, peptide:N-glycosidase (from almond emulsion), which specifically releases N-linked oligosaccharides by cleaving between the first internal GlcNAc residue and the asparaginyl moiety (41-43), 9.2, 21.3, and 49.5% of the label was released as free oligosaccharides from B-lymphocytes, T-lymphocytes, and thymocytes, respectively (Fig. 8). In fact, the quantitative agreement between Kinetics of @-elimination of pronase glycopeptides from exogalactosylated splenic lymphocytes. Splenic lymphocytes were isolated, exogalactosylated, and acid-insoluble proteins were digested exhaustively with predigested pronase. The limit pronase glycopeptides were isolated on Bio-Gel P-4 and the major larger species were then pooled, subjected to standard @elimination conditions for increasing lengths of time, and the mixtures were analyzed by Bio-Gel P-4 chromatography. Only the larger glycopeptides were used in order to not mask the appearance of any possible small intermediates (compare with Fig. 4, miniprint). The peak of material released by /"mination exactly coelutes with Gal,61-4GlcNAcitol. these results and those involving dilute alkali to determine oligosaccharide-protein linkage is quite striking. Also, when tryptic glycopeptides of exogalactosylated proteins from Tlymphocytes or B-lymphocytes were treated with the peptide:N-glycosidase and chromatographed on high resolution reverse phase HPLC (41,53), there was almost no detectable difference in the complex tryptic patterns as compared with untreated tryptic glycopeptides, indicating that very little of the radioactivity was released as free oligosaccharide (data not shown). Since the peptide:N-glycosidase quantitatively releases N-linked saccharides from every one of the several lymphocyte N-linked glycoproteins which we have examined (41,53), these results strongly suggest that the added galactose is not part of an unusual glycosidic linkage involving terminal GlcNAc in N-linked oligosaccharides. It should be noted that peptide:N-glycosidase will also cleave a single unsubstituted GlcNAc-asparaginyl linkage in a glycopeptide, such as that left by the action of endoglycosidase H (42). Taken together, the @-elimination and glycosidase data clearly indicate that a greater proportion of N-linked oligosaccharides are exogalactosylated in thymocytes than in either of the two peripheral lymphocyte types. Furthermore, the results indicate that the bulk of the ['HH]galactose incorporated into proteins by exogenous galactosylation is 0-glycosidically linked and found on a disaccharide the size of Galpl-4GlcNAcitol after @-elimination under reducing conditions. Identification of the @-Eliminated, Exogalactosylated Disaccharide-In order to identify the 3H-galactosylated disaccha- ride, the peaks from @-eliminated samples, which had been chromatographed over Bio-Gel P-4, were pooled, lyophilized, and analyzed by HPLC on a styrene-based, sulfonic acid column (0.5 X 100 cm) in the calcium form (Fig. 9). These columns are capable of extremely high resolution of low molecular weight, neutral oligosaccharides. As shown in Fig.  9A, authentic disaccharides, GalPl-4GalNAcito1, Gal@l-4GlcNAcito1, and Galpl-3GalNAcito1, which differ only in the position of a single hydroxyl moiety or linkage, are well resolved by this type of chromatography. The disaccharide obtained from @-elimination of exogalactosylated lymphocytes (all three cell types) exactly coelutes with the GalPI-4GlcNAcitol standard, both by comparison of radiolabeled standards in separate runs and by co-chromatography of the standard and the unknowns, usingultraviolet absorbance (210 nm) to detect the standard disaccharide. HPLC analyses of the disaccharides on an amino-bonded column, which separates neutral saccharides by hydrogen-bonding interactions, also demonstrates that the unknown disaccharides from all three lymphocyte types exactly coelutes with authentic GalP1-4GlcNAcitol (data not shown). Treatment of the labeled proteins or oligosaccharides with highly purified P-galactosidase (Escherichia coli) indicated that most of the added galactose was terminal and of the correct anomeric configuration (69% released).
internal and external galactosylatable GlcNAc residues, and also to directly determine the identity of the exogalactosylated hexosamine, splenic lymphocytes (a mixture of B-and Tlymphocytes) were preincubated with [3HH]glucosamine and subsequently exogalactosylated using unlabeled UDP-galactose. In companion experiments, prelabeled cells were permeabilized with detergent (0.25% NP-40) just prior to galactosylation with UDP-['4C]galactose. Fig. 10 shows the results of Bio-Gel P-4 (-400) chromatography of these galactosylated, [3HH]glucosamine-labeled proteins which have been subjected to &elimination in the presence of excess NaBH4. Table  1 summarizes several such experiments in quantitative terms. Several conclusions are evident from these data. 1) From 5-11% of the total [3H]glucosamine incorporated by splenic lymphocytes during a 12-h labeling is found in material the size of an N-acetylhexosaminitol monosaccharide after 0elimination. Compositional analysis, in fact, demonstrates that at least 61% of this material contains N -a~e t y l [~H ] galactosaminitol. 2) @-Eliminated disaccharide accounts for only 1-2% of the incorporated [3H]gluc~~amine radioactivity Jmphocyte Cell Surfaces 3313 in cells which have not been galactosylated. 3) In intact lymphocytes which have been exogalactosylated, only a small percentage (8%) of the 6-eliminatable [3H]hexosamine radioactivity is converted to disaccharide. 4) However, if the cells are first permeabilized with detergent, 60% of the p-eliminatable [3H]hexosamine is converted to disaccharide by galactosylation (Fig. 10d). As shown in Fig. 11 (miniprint), HPLC analysis confirms the proportions of this material which are converted to the disaccharide, Gal(31-4[3H]GlcNAcitol. Direct compositional analysis of the hydrolyzed, doubly labeled material also confirm the identity of the disaccharide. These findings provide direct evidence that many cell-surface and internal proteins of lymphocytes contain one or more 0glycosidically linked GlcNAc monosaccharide residues. Furthermore, the bulk of these galactosylatable terminal GlcNAc residues are either cryptic or found on the inside of the cell.

DISCUSSION
The aim of this study was to establish the conditions for using bovine milk galactosyltransferase as an impermeant probe for terminal GlcNAc residues on the surfaces of various types of lymphocyte subpopulations. Our plan was to not only quantify and identify proteins containing these residues, but also to characterize the nature of the oligosaccharides to which they are attached. Previous studies have used bovine milk galactosyltransferase as a probe for terminal GlcNAc on isolated saccharides, microsomal vesicles, or cell surfaces (54)(55)(56)(57). Analysis of galactosylated cell-surface glycoproteins, in many of the earlier studies on cell-surfaces, was performed by SDS-PAGE, isoelectric focusing, or a combination of both. However, the nature of the cell-surface exogalactosylated oligosaccharides has not been previously examined for any cell type.
Results described in this paper indicate the following. 1) Galactosyltransferase can readily be used to probe for the accessibility of terminal GlcNAc residues on the surfaces of living lymphocytes, under conditions where virtually all of the incorporated [3H]galactose is due to vectorial addition at the cell surface. 2) Functionally different subpopulations of lymphocytes differ greatly in numbers of exogalactosylatable GlcNAc moieties. 3) Thymocytes contain a much greater proportion of their accessible, terminal GlcNAc residues on N-linked oligosaccharides than do T-lymphocytes or B-lymphocytes from the same animal. 4) Numerous proteins at the cell surfaces of lymphocytes appear to contain one or more 0glycosidically linked GlcNAc monosaccharide moieties, which are readily exogalactosylatable in living cells. However, the bulk of the galactosylatable GlcNAc residues are cryptic or localized internally.
Evidence from at least eight independent lines of investigation indicate that we are, in fact, looking at vectorial and not metabolic incorporation of galactose by these exogalactosylation procedures. Perhaps the most convincing is that a 1000-fold excess of unlabeled galactose has no effect on exogalactosylation, and that virtually none of the incorporated radioactivity is epimerized to other monosaccharides, as it certainly would be if it were derived from metabolic routes (23). Many of the arguments employed here have been used previously in studies of endogenous cell-surface glycosyltransferases (58), in which the problem of establishing vectorial surface labeling is much more difficult than it is for exogenously added probes. Autoradiographic analyses of exogalactosylated lymphocytes have also directly eliminated the possibility that we might be labeling a small population of lysed or dead cells.
The finding of only 3% of the incorporated galactose in the Terminal GlcNAc on Lymphocyte Cell Surfaces lipid fractions has been observed by others using this approach in other cell types (54). Our extraction experiments suggest that most of the exogalactosylated proteins, which are cell associated after washing under isotonic conditions, are integral membrane proteins. Nearly all of the important lymphocyte antigens are also integral membrane components (1).
The quantitative differences in the extent of exogalactosylation observed among the different functional phenotypes of lymphocytes undoubtedly reflects the more general differences that exist in their overall cell-surface saccharide topography, among even morphologically identical, but functionally distinct, cells. Many studies involving lectins (10, 15), galactose oxidase (13, 21), or physical separation techniques (16, 20), have suggested that the cell-surface saccharide topography of lymphocytes correlates with their functional phenotypes. Early studies have suggested that these cell-surface saccharides are important for lymphocyte homing to their correct location in the peripheral lymphoid organs (59,60). More recent studies have suggested that the complex saccharide structures on lymphoid histocompatibility antigens are involved in their functions at the cellular level (8,9). Numer-ous developmental, stage-specific differences in the glycosylation of lymphocyte cell-surface proteins have also been observed (10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20)(21). It is clear that the significance of many of these observed developmental differences must await structural-functional (developmental) analyses at the level of single glycosylation sites on homogeneously purified differentiation antigens (41,53). In any case, the approach taken in this study does provide a quantitative estimate of the numbers and distribution of terminal GlcNAc residues on the surfaces of lymphocytes of functionally different phenotypes.
We have also attempted to characterize the nature of the oligosaccharide moieties to which the exogenously added galactose is attached. The finding that the bulk of the [3H] galactose was attached 0-glycosidically to protein and found on the disaccharide, Gal@l-4GlcNAcitol, after @-elimination was, at first, surprising. The possibility of proteins containing one o r more GlcNAc monosaccharides, especially ones which are 0-glycosidically linked, has not previously been described. Insensitivity of both the exogalactosylated proteins and their tryptic fragments to digestion with the peptide:N-glycosidase support these @-elimination results. More importantly, they also rule out the possibility that the observed galactosylated disaccharide results from alkali-catalyzed cleavage of an unusual glycosidic linkage involving terminal GlcNAc on Nlinked oligosaccharides.
Another possibility, namely that the &elimination conditions might be quantitatively cleaving an internal glycosidic linkage involving an exogalactosylated, nonreducing terminal GlcNAc on a typical 0-linked oligosaccharide, seems unlikely based upon kinetic analyses of the &elimination of galactosylated intact proteins and limit pronase glycopeptides. These kinetic studies indicated that no intermediate oligosaccharides larger than the disaccharide were released, as would be expected in the case of "peeling" (44) or endohydrolytic cleavage of a sugar-sugar glycosidic bond.
We have attempted to eliminate possible sources of artifact in these observations by the use of five independent criteria to establish that the exogalactosylated GlcNAc monosaccharides were originally covalently attached to lymphocyte proteins. The identification of the disaccharide, GalP1-4GlcNAcito1, which accounts for greater than 9075, greater than 80%, and at least 49% of the radioactivity from exogalactosylated and @-eliminated B-lymphocytes, T-lymphocytes, and thymocytes, respectively, is not only based upon the chromatographic and compositional criteria, discussed above, but also is consistent with the specificity of the galactosyltransferase employed (25)(26)(27)32). The existence of this apparently unusual monosaccharide structure on many lymphocyte surface proteins and the fact that thymocytes have a much higher proportion of N-linked glycoproteins with terminal, accessible GlcNAc residues may prove to be biologically significant. Even though the bulk of the exogalactosylation is on the 0-linked structures, the levels of incorporation are such that it should be possible to examine the N-linked oligosaccharides in more detail, even from B-lymphocytes.
The 0-linked GlcNAc monosaccharides may not have been observed in previous metabolic labeling studies because they appear to account for only a small percentage of the incorporated radioactivity and, after &elimination, they could be easily taken for small amounts of contaminating unincorporated radioactivity or degradation products. If the GlcNAc is attached by the hydroxyl moiety of serine (threonine), and only a fraction of a particular hydroxyamino acid residue in a well studied protein were glucosaminylated, the presence of the GlcNAc could be easily undetected, especially considering the difficulties associated with sequencing at serine or threonine residues. Alternatively, the 0-linked GlcNAc residues may not be present on previously well studied glycoproteins.
Earlier studies have also detected single GlcNAc moieties attached to proteins. Hase et al. (61) studied the carbohydrate heterogeneity of Taka-amylase and found a fraction which contained a single GlcNAc as the sugar moiety. These workers concluded that this moiety was the result of endoglycosidase action. Clearly, based upon the peptide:N-glycosidase and pelimination data, we have eliminated the possibility that the exogalactosylated GlcNAc moieties on lymphocytes arose from the action of either endogenous or exogenous endohexosaminidases acting on N-linked oligosaccharides.
Several very elegant studies have demonstrated the value of highly purified glycosyltransferases in structural-functional problems of receptor function (22) and as probes of saccharide accessibility (56). As vectorial probes of cell-surface saccharide topography, glycosyltransferases have many advantages which make them valuable tools in addition to the conventional approaches using lectin binding. 1) Transferases are extremely specific not only with respect to binding, but also with respect to the reaction they catalyze. 2 ) Both the donor substrates and the transferase are impermeable to the plasma membrane. 3) Unlike lectins, the glycosyltransferases cause covalent specific modification of the saccharides they interact with, thus allowing radiolabeling, isolation, and characterization of the vectorially labeled glycoconjugates. 4) Covalent modification by exogenous glycosyltransferases allows specific questions of turnover, modification, and saccharide function to be addressed, which would be difficult or impossible by other approaches.
Detailed analyses of oligosaccharide structural microheterogeneity at individual glycosylation sites on important homogeneously purified antigens, which involves the use of metabolic radiolabeling (41, 53), is nicely complemented by similar approaches examining these same glycosylation sites, which have been vectorially radiolabeled at the cell surface. This latter approach has the potential of more accurately detecting subtle topographical alterations or localizations of particular saccharide moieties at the cell surface actually seen by interacting cells, and results in covalent labeling of only those saccharide moieties which are exposed.