Brain Contains HNK-1 Immunoreactive O -Glycans of the Sulfoglucuronyl Lactosamine Series that Terminate in 2-Linked or 2,6-Linked Hexose (Mannose)*

The monoclonal antibody HNK-1 originally raised to an antigenic marker of natural killer cells also binds to selected regions in nervous tissue. The antigen is a car- bohydrate that has attracted much interest as its ex-pression is developmentally regulated in nervous tissue, and it is found, and proposed to be a ligand, on several of the adhesive glycoproteins of the nervous system. It is also expressed on glycolipids and proteoglycans, and is the target of monoclonal auto-antibodies that give rise to a demyelinating disease. The epitope, as characterized on glycolipids isolated from the nervous system, is expressed on 3-sulfated glucuronic acid joined by (cid:98) 1-3-linkage to a neolacto backbone. Here we exploit the neoglycolipid technology, in conjunction with immuno-detection and in situ liquid secondary ion mass spec- trometry, to characterize HNK-1-positive oligosaccharide chains derived by reductive alkaline release from total brain glycopeptides. The immunoreactive oligosaccharides detected are tetra- to octasaccharides that are very minor components among a heterogeneous population, each representing less than 0.1% of the starting material. Their peripheral and backbone sequences re-semble those of the HNK-1-positive glycolipids. An un- incor- porating a mild L (DPPE) the disaccharide alditol Man (cid:97) 1–2Manol. This was differentiated from -3Manol, -2Galol, and -2Glcol prepared from Man (cid:97) 1–3Manol, Fuc (cid:97) 1–2Galol, and Glc (cid:98) 1– 2Glcol, respectively; the retention times being 10.03, 9.81, 10.07, and 10.32 min for partially methylated alditol acetate of -3Manol, -2Glcol, -2Manol, and -2Galol, respectively. Antibody Binding Assays— Hybridoma cells secreting the HNK-1 monoclonal antibody (7) were obtained from the American Type Culture Collection (Rockville, MD), and antibody containing ascites fluid was produced by standard procedures following inoculation BALB/c mice. was assayed as described (40), using a 1:500 dilution of ascites overlaid at room temperature for 2 h. 3-Sulfoglucuronyl neolactotetraosylceram-ide (HNK-1 glycolipid) isolated from bovine cauda equina (a gift of Dr. F. B. Jungalwala; Eunice Kennedy Shriver Center, Waltham, MA) was used as a positive control. Linkage and Composition Analyses of the Oligosaccharide Alditol 1A2c— Monosaccharide composition and linkage information on the major component in 1A2c was obtained by methylation analysis of the free oligosaccharide alditol preparation. The results showed -3GlcA1-, -3Gal1-, -4GlcNAc1-, and -2Ma-nol, in the form of the partially methylated alditol acetates 2,4-di- O -methylglucitol, 2,4,6-tri- O -methylgalactitol, 3,6-di- O -methyl-2- N -methyl-acetamidoglucitol, and 1,3,4,5,6-penta- O -methylmannitol, respectively. With these and the mass spectrometric data the main oligosaccharide in fraction 1A2c is assigned as: HSO 3 -3GlcA1–3Gal1–4GlcNAc1–2Manol. TLC-LSIMS Immunoreactive 2A2d 3A2f

Monoclonal antibodies raised with the aim of identifying developmentally regulated antigens and differentiation antigens that have biological functions are in many cases directed to oligosaccharides of glycoproteins or glycolipids (1). Prominent among such antigens are those based on backbone sequences of the poly-N-acetyllactosamine type that are O-or N-glycosidically linked to proteins or to a glucose core of glycolipids (1,2), several of which are now established ligands for carbohydrate-binding proteins of animals (3)(4)(5)(6); and there is much interest in the biological events elicited by the carbohydrate-protein interactions. HNK-1 antigen, recognized by a hybridoma antibody, is a carbohydrate antigen first identified as a marker of the natural killer cell population among lymphocytes (7), and later found to be expressed on glycolipids (8 -10) as well as several glycoproteins (11) and proteoglycans (12) of the nervous system, with a changing distribution at different stages of development (13)(14)(15). HNK-1 antigen, as characterized biochemically on glycolipids, is expressed on 3-sulfated glucuronic acid joined by ␤1-3-linkage to a (poly)Nacetyllactosamine backbone (8 -10). This carbohydrate sequence has been implicated as a ligand for the leukocyteendothelium adhesion molecules, L-and P-selectins (16,17) and for a number of cell-interaction systems in nervous tissue (11,18), notably as a ligand for the major glycoprotein of myelin, P o , in the peripheral nervous system (19), the extracellular matrix protein, laminin (20), and an adhesive protein, amphoterin, of the cerebellum (21). Earlier indications of the presence of the HNK-1 antigen on N-glycosidically linked oligosaccharides (22)(23)(24) have now been followed by structural studies demonstrating the presence of the 3-sulfoglucuronyl N-acetyllactosamine sequence on a N-glycan isolated from glycoprotein P o (25). Biosynthetic experiments suggest that different glucuronyltransferases are involved in the synthesis of the HNK-1 epitope on glycolipids and glycoproteins (26).
The present investigation is concerned with HNK-1-immunoreactive oligosaccharides among total brain glycopeptides. In exploratory experiments, oligosaccharides were released by hydrazinolysis to obtain N-glycosidically linked chains, and by reductive and non-reductive alkaline hydrolysis to obtain Oglycosidically linked chains; these were examined for immunoreactivities after conversion to neoglycolipids. Having observed immunoreactive oligosaccharides among all three oligosaccharide populations, a more detailed investigation has been performed of oligosaccharides released by reductive alkaline hydrolysis.

MATERIALS AND METHODS
Glycopeptides-Glycopeptides were prepared from a Pronase digest of a lipid-free protein residue (chloroform-methanol extract) of rabbit brain, after precipitation of glycosaminoglycans with cetylpyridium chloride (27).
Oligosaccharides-In small scale experiments, oligosaccharides were obtained from 10-mg aliquots of brain glycopeptides by three methods. (i) O-Glycosidic oligosaccharides in reducing form were obtained after subjecting the glycopeptides to hydrolysis with ethylamine (70% w/v in water) at 60°C for 6 h as described (28). The oligosaccharide-rich fraction was N-acetylated using acetic anhydride (29) to acetylate free amino groups of any residual peptide materials that might interfere with the reductive amination reaction in the generation of neoglycolipids. (ii) O-Glycosidic oligosaccharides in reduced form were obtained after subjecting the glycopeptides to hydrolysis with 0.05 M NaOH, 1 M NaBH 4 at 45°C for 16 h as described (30), and the oligosaccharide-rich fraction was N-acetylated (29). (iii) N-Glycosidic oligosaccharides in reducing form were obtained after treatment of the glycopeptides with * This work was supported by Medical Research Council Program Grant E 400/622, Arthritis and Rheumatism Council Project Grant F0061, and National Institutes of Health Grants NS-13876 and MH-0012. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. anhydrous hydrazine for 9 h (29). In a larger scale experiment, Oglycosidic oligosaccharides in reduced state were obtained from 600-mg glycopeptides after reductive alkaline hydrolysis (30).
Fractionation of Oligosaccharides-Oligosaccharides obtained in each of the small scale experiments were fractionated on DEAE Sephadex A-25 (0.75 ml, acetate form, Pharmacia) into neutral fractions, 1.5 ml (unretained with 10 mM NH 4 OAc, pH 5) and acidic fractions, 3 ml (eluted with 500 mM NH 4 OAc, pH 5); these were designated as N and A, respectively. Fractions A were treated with Arthrobacter ureafaciens sialidase (Boehringer Mannheim) and further fractionated as above into neutral (N s ) and sialidase-resistant acidic fractions: A s 1, and A s 2 eluted with 100 and 500 mM NH 4 OAc, respectively. Oligosaccharides recovered are summarized in Table I. An aliquot of oligosaccharides A s 2 (26 g of hexose) from the ethylamine hydrolysis experiment was further fractionated by isocratic elution on a normal phase HPLC 1 column (TSK amide-80, 4.6 ϫ 250 mm, Anachem, United Kingdom) using 70% acetonitrile/water in 10 mM sodium acetate, pH 5.3, at a flow rate of 1 ml/min. Eight fractions (1.5 ml each), designated A s 2 a-h, were collected after 2 min of flow, and desalted using AG 50W X-12 H ϩ (Bio-Rad); whole fractions were converted to neoglycolipids, and 10% aliquots of the neoglycolipid reaction mixtures tested for HNK-1 immunoreactivity.
Saccharides released in the large scale reductive alkaline hydrolysis experiment were initially resolved by two chromatographic steps using Bio-Gel columns (1.5 ϫ 100 cm; Bio-Rad) essentially as described (33) except that 2 mM trifluoroacetic acid was used as eluent at ambient temperature. Bio-Gel P-6 was used to separate high molecular weight materials located at the void volume from the smaller sized retained fractions, eluting between 61 and 150 ml. The retained fraction, 80.3 mg of hexose, was concentrated by evaporation and fractionated on Bio-Gel P-4. Seven pooled fractions were obtained ( Fig. 1): 1) 0.5 mg; 2) 3.1 mg; 3) 5.5 mg; 4) 3.6 mg; 5) 13.4 mg; 6) 52.0 mg; and 7) 1.7 mg, determined as hexose. Aliquots of these pooled fractions (5 g of hexose) were converted to neoglycolipids and tested for HNK-1 immunoreactivity (Fig. 2). The pooled Bio-Gel P-4 fractions 1, 2, and 3 which showed strong HNK-1 immunoreactivities were chromatographed on a weak anion exchange polymeric HPLC column (Asahipak NH 2 -P50, 4.6 ϫ 250 mm, Prolabo, United Kingdom) using a start solvent: 2.5 mM NH 4 OAc, pH 6.8, and eluting with 0 -250 mM NH 4 OH, pH 11.7 (Fig. 3). A neutral fraction eluting with the start solvent and two acidic fractions eluting at about 80 and 150 mM NH 4 OH were obtained, and designated N, A1, and A2, respectively (for example 1N, 1A1, and 1A2 for the three subfractions of Bio-Gel P-4 pool 1). Aliquots of these fractions (5 g of hexose) were converted to neoglycolipids and tested for HNK-1 immunoreactivities as neoglycolipids (results not shown). In each set, fraction A2 showed HNK-1 immunoreactivity and was further fractionated by gradient elution on a normal phase TSK amide-80 HPLC column as above using acetonitrile/water in 2 mM trifluoroacetic acid (Fig. 4). Aliquots of these chromatographic fractions (1 g of hexose) were monitored for HNK-1 immunoreactivities as neoglycolipids. From each of the A2 oligosaccharide fractions, three HNK-1 immunoreactive subfractions were obtained (Fig. 4): 1Ab, 1Ac, and 1Ad; 2Ac, 2Ad, and 2Ae; and 3Ad, 3Ae, and 3Af; the oligosaccharides recovered in terms of micrograms of hexose are given in Table II.
Hexose Assay-The hexose contents of oligosaccharides were estimated by a micro-scale procedure on silica gel plates by applying as spots onto the plates 1 l of test samples and galactose standards at a Twenty-six g of oligosaccharide fraction A s 2 (determined as hexose) from the ethylamine experiment was subfractionated into eight fractions (a-h) by normal phase HPLC as described under "Materials and Methods"; fraction g with the strongest HNK-1 immunoreactivity as neoglycolipids was subjected to TLC-LSIMS.
FIG. 1. Bio-Gel P-4 chromatography profile of saccharides released by reductive alkaline hydrolysis of brain glycopeptides in the large scale experiment. Oligosaccharides in the retained volume after Bio-Gel P-6 chromatography were chromatographed on a Bio-Gel P-4 column using 2 mM trifluoroacetic acid as solvent. Elution of oligosaccharides (900 l/fraction) was monitored by assays of hexose concentration. The numbered arrows indicate positions of glucose oligomer standards. Seven pooled fractions were obtained as indicated; shading indicates HNK-1 immunoreactivity detected in neoglycolipids prepared from aliquots of the pooled fractions (see Fig. 2).

FIG. 2.
Antibody binding assay to reveal HNK-1 immunoreactivity among neoglycolipids prepared from the pooled oligosaccharide alditol fractions 1 to 7 obtained in the large scale experiment. The neoglycolipid reaction mixtures derived from the Bio-Gel P-4 fractionated oligosaccharides containing 1 g of hexose were chromatographed in solvent B, revealed by primulin stain (panel p), and thereafter, overlaid with monoclonal antibody HNK-1, and binding detected with 125 I-labeled anti-mouse immunoglobulins and autoradiography (panel h) as described under "Materials and Methods." Arrow indicates positions of sample applications. 0.05-1.0 mg/ml, and staining with orcinol reagent (34).
Liquid Secondary Ion Mass Spectrometry (LSIMS) and TLC/LSIMS Analyses-LSIMS analyses were carried out as described previously (35,36). Native oligosaccharides (ϳ1 g) were analyzed after applying onto the target using thioglycerol as liquid matrix and neoglycolipids were analyzed directly on chromatograms after TLC.
Methylation Analysis-For methylation analysis of oligosaccharides, permethylation (37), hydrolysis, reduction, acetylation, and gas chromatography-mass spectrometry analysis were performed as described previously (38). A modification was introduced into the procedure to include reduction of carboxymethyl groups (39) after permethylation to detect uronic acid residues as their hexitols.
Antibody Binding Assays-Hybridoma cells secreting the HNK-1 monoclonal antibody (7) were obtained from the American Type Culture Collection (Rockville, MD), and antibody containing ascites fluid was produced by standard procedures following inoculation of BALB/c mice. Antibody binding to neoglycolipids chromatographed on HPTLC plates was assayed as described (40), using a 1:500 dilution of ascites overlaid at room temperature for 2 h. 3-Sulfoglucuronyl neolactotetraosylceramide (HNK-1 glycolipid) isolated from bovine cauda equina (a gift of Dr. F. B. Jungalwala; Eunice Kennedy Shriver Center, Waltham, MA) was used as a positive control.

Small Scale Experiments
In antibody binding experiments using the neoglycolipids obtained in the small scale experiments, HNK-1 immunoreactivities were observed predominantly in the most acidic fractions, A s 2 (Fig. 5), and not in the neutral fractions N and N s (not shown). The binding was to minor components which showed negligible primulin staining (not shown). Among the neoglycolipids derived from the reducing oligosaccharides released by hydrolysis with ethylamine, multiple immunoreactive components were detected, chromatographing in the tri-/ tetrasaccharide and higher oligomer regions. In the same chromatogram binding experiment, using neoglycolipids derived from acidic oligosaccharides released by hydrazinolysis, immunoreactivity was observed just above the point of application (this is the position of migration of neoglycolipids derived from multi-antennary N-glycosidic oligosaccharides), 2 FIG. 3. Anion exchange HPLC elution profiles of three of the immunoreactive oligosaccharide alditol pools 1, 2, and 3 in the large scale experiment. Oligosaccharide alditol pools 1, 2, and 3: 400, 3100, and 5500 g of hexose, respectively, obtained after Bio-Gel P-4 chromatography, were subjected to HPLC on an anion exchange polymeric column. Fractions were pooled as indicated (designated N, A1, and A2). Elution profile for pool 1 is that of a single chromatography run that was performed with this pool; profiles for pools 2 and 3 are representative traces of the elution profiles of one of four runs for each pool.
FIG. 4. Normal phase HPLC elution profiles of acidic A2 fractions of the oligosaccharide alditol pools 1, 2, and 3. The HNK-1 immunoreactive acidic oligosaccharide alditol fractions 1A2, 2A2, and 3A2: 220, 262, and 880 g of hexose, respectively, were subjected to normal phase HPLC and fractions were pooled as indicated; the entire subfraction a of 1A2, and 1 g (hexose) of the other subfractions were converted to neoglycolipids, and assayed for HNK-1 immunoreactivity as shown in Fig. 6. Pooled fractions a, as well as a part of pooled fraction b of 3A2 are not shown. Black boxes indicate strong HNK-1 immunoreactivity, striped boxes show medium immunoreactivity, and dotted boxes indicate weak immunoreactivity. and at a position corresponding to tri-/tetrasaccharide neoglycolipids. The hydrazine-released oligosaccharides were not investigated further in the present study.
TLC-LSIMS analyses of the HNK-1 immunoreactive neoglycolipids derived from subfraction g of fraction A s 2 (Table I)  , consistent with the compositions HSO 3 ⅐HexA⅐Hex 2 ⅐Hex NAc⅐DPPE and HSO 3 ⅐HexA⅐Hex 3 ⅐HexNAc 2 ⅐DPPE, respectively (results not shown). These findings, with the evidence in each case for the presence of an additional hexose, suggested the presence of the HNK-1 epitope on oligosaccharides having reducing-end hexose residues that may have been generated by peeling of O-glycosidic oligosaccharides. Having detected HNK-1 positive components among oligosaccharides in the small scale reductive experiment (Fig. 5), a larger scale alkaline borohydride reaction was undertaken to minimize peeling.

Large Scale Experiments
General Features of the Immunoreactive Oligosaccharide Alditols-When converted to neoglycolipids after periodate oxidation, the oligosaccharide alditol fractions 1A2c and d yielded a triplet of strongly immunoreactive components i, ii, and iii and a weakly immunoreactive component, iv (Fig. 6). Fractions 2A2c, d, and e and 3A2d, e, and f also yielded immunoreactive components at positions i and ii, but in addition there were components with decreasing mobilities v, vi, and vii in fraction 2A2, and vii, viii, and ix in 3A2 (Fig. 6).
The acidic oligosaccharide alditol fractions 1A2c, 2A2d, and 3A2f with the strongest HNK-1 immunoreactivities as neoglycolipids (Figs. 4 and 6) were selected for negative ion LSIMS analyses as the free oligosaccharides. They were deduced to contain sulfated tetra-to octasaccharides with uronic acid in five of seven of them (Table III) (Fig. 7) was consistent with that of a reduced tetrasaccharide HSO 3 ⅐HexA⅐Hex 2 ⅐HexNAc. This composition is the same as that of one of the components detected as a neoglycolipid (at position a, Fig. 5) among ethylamine-released oligosaccharides. Oligosaccharide analogs of 1A2c containing an additional HexNAc and additional Hex⅐HexNAc were deduced to be in fraction 2A2d; the latter corresponded to the immunoreactive component detected (at position b, Fig. 5) in the ethylamine experiment. Analogs of 1A2c with additional Hex 2 ⅐HexNAc 2 and dHex⅐Hex 2 ⅐HexNAc were deduced to be in fraction 3A2f and the other oligosaccharide alditols to contain sulfate, dHex⅐Hex 3 ⅐HexNAc (fraction 2A2d) and dHex⅐Hex 4 ⅐ HexNAc 2 (fraction 3A2f). On account of its relative homogeneity, fraction 1A2c was investigated in the greatest detail as follows.
TLC-LSIMS Analyses of the Immunoreactive 1A2c Neoglycolipids-The 1A2c-derived triplet of immunoreactive neoglycolipid bands i, ii, and iii (Fig. 6) Fig. 7 for band ii). Band iv gave too weak a spectrum for interpretation. The 30-Da differences in the molecular ions derived from bands i to iii are indicative of increments of CHOH arising from different positions of periodate cleavage of a terminal hexitol residue at consecutive C-C bonds (41). This is illustrated for 2-substituted mannitol 3 in Scheme 1, where R ϭ oligosaccharide and the resulting neoglycolipids contain remnants of the reduced hexitol with three, four, or five carbons. On this basis, and the deduced composition of the free, non-oxidized oligosaccharide alditol 1A2c (Table III), the [M Ϫ H] Ϫ ions of the triplet of neoglycolipids were interpreted to arise from neoglycolipids with the composition HSO 3 ⅐HexA⅐Hex⅐HexNAc-X-DHPE, where X ϭ the hexitol-derived fragments, C 3 H 5 OH (band i), C 4 H 6 (OH) 2 (band ii), and C 5 H 7 (OH) 3
TLC-LSIMS Analyses of the Immunoreactive 2A2d and 3A2f Neoglycolipids-Of the immunoreactive neoglycolipid bands derived from 2A2d and 3A2f, the bands at i/ii position (Fig. 6) gave the same [M Ϫ H] Ϫ ion, m/z 1357, as that from 1A2c band i with the three carbon-containing hexitol remnant (Table IV). A small ion was also present at m/z 1327 corresponding to an analog with a two carbon-containing remnant. The presence of the tetrasaccharide-derived neoglycolipids at positions i/ii originating from alditol fractions 2A2 and 3A2 are unlikely to reflect an incomplete separation of the tetrasaccharide in the adjacent fractions 1, 2, and 3 during the Bio-Gel P-4 chromatography step. A more likely explanation is that a proportion of the 2A2 and 3A2 alditols are branched at the terminal hexitol, and the periodate oxidation cleavage is limited to two rather than three C-C bonds as shown in Scheme 2 for 2,6-disubsti-tuted mannitol, 4 where R and RЈ are oligosaccharides.
Thus the mass spectra indicated that the bands at i/ii positions were derived by a cleavage of a 2,6-branched hexitol (cf. Scheme 2) with the HSO 3 ⅐HexA⅐Hex⅐HexNAc sequence linked at the 6-position. The neutral neoglycolipids that would be predicted to arise from the 2-linked branch were not specifically searched for in the present study by scanning of the di-to tetrasaccharide regions. However, the compositions of the nonoxidized penta-, hexa-, and octasaccharide alditols with [M Ϫ H] Ϫ ions 1005, 1167, 1458, and 1532 (Table III) are consistent not only with unbranched 2-hexitol-terminating sulfated sequences, but also with the presence of non-sulfated outer arms consisting of HexNAc, Hex⅐HexNAc, dHex⅐Hex⅐HexNAc, and Hex 2 ⅐HexNAc 2 , respectively, 2-linked to terminal hexitols that also bear 6-linked sulfated outer arms.
Of the slower migrating triplet of bands derived from 2A2d, band v gave too weak a spectrum for interpretation, while bands vi and vii gave [M Ϫ H] Ϫ ions at m/z 1752 and 1782, respectively (Table IV). These correspond to higher molecular mass analogs of neoglycolipid bands ii and iii derived from 1A2c, but 365 Da higher, indicating in each case, the presence of an additional disaccharide unit, Hex⅐HexNAc. Fragment ions m/z 1672 and 1496 from band vi showed a terminal sulfated HexA residue, and thus provided partial sequence HSO 3 -HexA. Bands vi and vii are deduced to have been derived from a hexitol-terminating polylactosamine-type analog of 1A2c.
Of the slowest migrating duplex of immunoreactive bands viii and ix from 3A2f, the former gave an [M Ϫ H] Ϫ ion at m/z 1898, and fragment ions m/z 1818 and 1642 which indicated the presence of a sulfated terminal HexA residue and provided composition information and the partial sequence HSO 3 ⅐HexA- (Table IV). An [M Ϫ H] Ϫ ion at m/z 1868 was also present in this band and allowed assignment of composition. Band ix gave two [M Ϫ H] Ϫ ions at m/z 2117 and 2147; these are 365 Da higher than those from bands vi and vii, respectively, and consistent in each case, with the presence of an additional lactosamine-type disaccharide, Hex⅐HexNAc. Collectively, these data with 2A2d and 3A2f neoglycolipids, together with the mass spectra from the parent non-oxidized oligosaccharide 4 A 2,4-disubstituted terminal hexitol is ruled out as it would be cleaved only between C5 and C6 to give a product containing both oligosaccharide substituents (R and RЈ), while a 3,6-disubstituted hexitol would be cleaved between C4 and C5 to give R-(C1-C4) and RЈ-(C5-C6), and is also ruled out.   7. Negative ion mass spectra of underivatized oligosaccharide alditol subfraction 1A2c and immunoreactive neoglycolipid band ii derived from 1A2c after mild periodate oxidation. Top panel shows the mass spectrum of the non-derivatized oligosaccharide alditol 1A2c, and the lower panel shows the mass spectrum of 1A2c neoglycolipid. An ion at m/z 1357 indicates the presence of a small amount of band i neoglycolipid containing the C 3 H 5 OH hexitol remnant (see Table III). alditols (Table III), indicate that these HNK-1 immunoreactive hexa-and octasaccharides are a population of polylactosaminetype glycans terminating in 2-substituted hexitol which upon oxidation is cleaved to produce remnants of the reduced hexitol with three, four, or five carbons (Scheme 1, Table IV). DISCUSSION This investigation illustrates the power of the neoglycolipid technology to detect and biochemically characterize bioactive oligosaccharides that are extremely minor components of a heterogeneous population. It has enabled total glycopeptidederived oligosaccharides from brain, a sample size of 80 mg of hexose, to be monitored for immunoreactivities during successive chromatographies, and provided critical information on sequence and linkage by in situ TLC-LSIMS analyses of discrete immunoreactive components that represent as little as 0.03% of the starting oligosaccharides.
The small scale experiments indicated that both N-and Oglycosidic oligosaccharides contain the HNK-1 determinant. The present study is focused on the immunoreactive oligosaccharides obtained by reductive alkaline hydrolysis, the procedure of choice for releasing O-glycosidic, N-acetylgalactosamine-terminating oligosaccharides with minimal peeling (30). The majority of these were found to be tetra-to octasac-charides. Their peripheral and backbone sequences resembled those of the HNK-1-positive glycosphingolipids isolated from nervous tissue (8,9) in being of the 3-sulfoglucuronyl lactosamine series. Among the oligosaccharides obtained under non-reductive conditions, using ethylamine or hydrazine, there was immunoreactivity of neoglycolipids chromatographing as trisaccharides. These are likely products of peeling as they were not detected among the oligosaccharides isolated under the reductive conditions (Fig. 4). There was evidence for the presence of a deoxyhexose residue in one of the immunoreactive hexasaccharides released from the brain glycopeptides. With the amounts available, however, the precise location of this presumed fucose residue along the oligosaccharide chain has not been established.
Not all of the sulfated oligosaccharides identified in the nonderivatized fractions 2A2d and 3A2f were uronic acid-containing (Table III): two of the deoxyhexose containing sulfated oligosaccharides, one in 2A2d and another in 3A2f, were lacking in uronic acid, and compositionally they resembled the oligosaccharides identified previously among oligosaccharides from brain chondroitin sulfate proteoglycans (42). These were not detected, however, among the immunoreactive neoglycolipids derived from fractions 2A2d and 3A2f.
In studies which will be described elsewhere, N-acetylgalactosaminitol-terminating oligosaccharides have been detected, as expected, by methylation analyses in Bio-Gel P-4 fractions 1 to 4. It is unexpected, however, to find that the immunoreactive oligosaccharides in fractions 1, 2, and 3 are hexitol-terminating, some 2-substituted and others 2,6-disubstituted. In oligosaccharide 1A2c the hexitol was identified as 2-substituted mannitol. In fraction 3, additional mannitol-terminating oligosaccharides Gal1-4GlcNAc1-2Manol and Gal1-4/3(Fuc1-3/ 4)GlcNAc1-2Manol have been identified after purification and methylation analyses (not shown). Moreover, a recent report describes an analog of the former oligosaccharide 3-sialylated at galactose among the oligosaccharides isolated from the Schwann cell glycoprotein ␣-dystroglycan (43). This raises important questions as to the origins of this family of hexitolterminating oligosaccharide alditols. Are they peeled products of N-glycosidic oligosaccharides? Or are they members of an unusual family of O-glycosidic oligosaccharides linked to protein via mannose residues, of the type previously described to occur among oligosaccharides released by reductive alkaline SCHEME 2  3 -DHPE a Sequences given by mass spectra are indicated by dashes; where sequence information was not obtained, the components are separated by periods. The set of hexitol fragments in bands i, ii and iii derived from oligosaccharide alditol 1A2c, depicted by -OC 3 H 5 OH, -OC 4 H 6 (OH) 2 , and -OC 5 H 7 (OH) 3 , respectively, are the predicted periodate oxidation products of 2-substituted hexitol (the same applies for fragments from 2A2d in bands vi and vii, and from 3A2f in bands viii and ix); fragments in bands at positions i/ii derived from alditols 2A2d and 3A2f, depicted by -OC 2 H 4 and -OC 3 H 5 OH, are predicted products of 2,6-disubstituted hexitol. b ND, not determined. SCHEME 1 degradation from glycopeptide domains of brain proteoglycans (44)? In the studies which will be described elsewhere, oligosaccharides of N-glycosidic type with intact chitobiosyl cores have been detected among the pooled Bio-Gel P-4 fractions; these are of high-mannose type in fractions 4 and 5 and complex type in fraction 5; these findings are consistent with evidence reviewed in Ref. 45 that at least 10 -20% of N-glycosidic oligosaccharides are released from protein after alkaline borohydride degradation. In the present investigation, however, no evidence has been found for the presence of peeled products of N-glycosidic oligosaccharides with loss of one or two of the N-acetylglucosamines of the chitobiosyl core, or other intermediate products of sequential degradation among the Bio-Gel P-4 fractions. Moreover, in separate investigations 2 when a purified, asialo-biantennary N-glycosidic oligosaccharide (100 g) was subjected to the reductive alkaline borohydride treatment, and the product mixture was subjected to HPLC on a TSK amide 80 column, no appreciable degradation product was detected, and greater than 98% of the starting material was recovered. Thus, unless extensive peeling of N-glycans is catalyzed, for unknown reasons, in the course of their release from the brain glycopeptides under the conditions of reductive alkaline borohydride degradation, the immunoreactive neoglycolipids detected in the present study are likely to represent O-glycans. The 2-and 2,6-disubstitution of the end-hexitol residues is established here with confidence for the immunoreactive oligosaccharides from the distinctive spectra of the neoglycolipids containing the hexitol remnants produced by oxidation of the end-hexitols. In the tetrasaccharide 1A2c, the end-monosaccharide is mannitol as shown by methylation analysis. Among the hexa-and octasaccharides, the presence of 2-substituted and 2,6-disubstituted end-hexitol is demonstrated, but there is no evidence for the presence of 3-substituted hexitol. These findings differ from the assignment of 3-substituted mannitol for the oligosaccharide alditols isolated from the glycopeptide moieties of brain chondroitin sulfate proteoglycans of the rat (44). In the earlier study, methylation analysis was performed on a tetrasaccharide alditol, but the partially methylated alditol acetate of Manol could not be visualized due to interfering peaks in GC; the inference of 3-substitution of the mannitol was made by comparing the mass spectra of the trimethylsilyl derivatives (and permethyl) of a disaccharide GlcNAc-Manol produced after treatments of the glycopeptides with alkaline NaBH 4 or alkaline NaBD 4 . A fragment at m/z 205 (m/z 89 for the permethylated derivative) containing the carbon atoms 1 and 2 or 5 and 6, and arising by cleavage of different C-C bonds, showed a 10% shift in intensity to m/z 206 in the deuterated derivatives. Although this is a predicted feature of a 1-3-linked disaccharide, we have observed a similar shift (13%) in intensity m/z 205 to 206 in a 1-2-linked disaccharide standard GlcNAc␤1-2Manol, generated by NaBH 4 and NaBD 4 treatments of the reducing disaccharide 2 ; this might represent a small degree of rearrangement involving the deuterium. We find that the spectrum of the deuterated disaccharide is very similar to that of the deuterated GlcNAc␤1-3Manol published earlier (44). It is desirable therefore to re-evaluate the end sequences of oligosaccharide alditols derived from the glycopeptide moieties of brain chondroitin sulfate proteoglycans. The biosynthesis of these mannose-terminating glycans is an important topic for future investigations, for there is a distinct possibility that they are products of the O-mannosylation pathway (46) described in fungi.