Txe JOURNAL OF BIOLOGICAL CHEMISTRY Voi. 26I, No. 4, Issue of February 5, pp. 1599-1608, I986 0 1986 by The American Society of Biological Chemists, Inc. Printed in U. 5’. A.

Characterization of the Oligosaccharide Alditols from Ovarian Cyst Mucin Glycoproteins of Blood Group A Using High Pressure Liquid Chromatography (HPLC) and High Field ‘N NMR Spectroscopy”

(Received for publication, April 22, 1985)

Virendra K. Dua$, B. N. Narasinga RaoS, Shing-Shing WuS, Volker E. Dubs$, and C. Allen Bush$ From the Wepartment of Chemistry, Illinois Institute of Technology, Chicago, Illinois 60616 and the $Department of Pathology, Euanstoon. Hospital, Euanston, Illinois 60201

A combination of reverse phase and normal phase high pressure liquid chromatography has been used to separate the reduced oligosaccharides produced by al- kaline borohydride degradation of a blood group A ovarian cyst mucin glycoproteins. Fourteen com- pounds, ranging in size from a monosaccharide to a decasaccharide, have been isolated preparatively using a Zorbax C-18 reverse phase column eluted with water and a MicroPak AX-5 normal phase column eluted with aqueous acetonitrile. The purity of the products and their structures were determined from the fully as- signed high field proton NMR spectra. The resonances of exchangeable amide protons, observed by the Red- field selective pulse sequence in WzO, were assigned by decoupling to the resonances of H2 of the 2-acetamido sugars. Nuclear Overhauser effects were used to estab- lish the relationship of the anomeric protons and those of the aglycone. In exception to earlier proposals that nuclear Overhauser effect on irradiation of the ano- meric proton should always be observed at the proton attached to the aglycone carbon, we find that for the linkage of GalNAcp( 1-*3)Gal, nuclear Overhauser ef- fect on irradiation of the a-anomeric proton resonance is observed not at H3 but at H4 of galactose. A combi- nation of NMR methods and enzymatic degradation was employed to determine the structures of 13 differ- ent oligosaccharides of which seven have not previ- ously been reported. These oligosaccharides, which terminate with &Gal, a-Fuc, 8-GlcNAc, and a-GalNAc, account for 75% of the total glycoprotein carbohy- drate, the remainder being isolated as a mixture of glycopeptides and a high molecular weight polysaccha- ride whose NMR spectrum implies a simple repeating subunit structure closely related to that of the oligo- saccharides.

Immunological and chemical investigations of the oligosac- charides of the mucin glycoproteins of ovarian cysts have identified the carbohydrate structures responsible for blood group activity (1). Based on the oligosaccharides isolated from blood group H and Lewis a and b (2, 3) and those from €3 active ovarian cyst glycoproteins (4, 5) a composite blood group megalosaccharide structure was formulated. The pe- ripheral portions of the oligosaccharides from a blood group

*This research was supported by National Institutes of Health Grant GM 31449 and National Science Foundation Grant DMB 821703. The costs of publication of this article were defrayed in part hy the payment of page charges. This article must therefore be hereby marked “ ~ u e ~ t ~ ~ e ~ e ~ ~ ” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

A glycoprotein, carrying the specific determinant N-acetyl- galactosamine, were identified by Kabat and co-worker (6) as products of peeling (base degradation reaction). More re- cently, Wu et ab. (7,8) have determined the structures of the core oligosaccharides derived by periodate oxidation of blood group A active ovarian cyst glycoproteins. However, the com- plete structures of oligosaccharides derived from blood group A ovarian cyst mucins have not been reported. In a recent report, Slomiany et al. (9) have isolated a number of blood group A active oligosaccharides from human gastric mucin and reported partial structures which appear to be closely related to those of A active ovarian cyst mucins.

High field proton NMR spectroscopy is a powerful method for structure determination of reduced oligosaccharides iso- lated by alkaline borohydride degradation of mucins. van Halbeek et ul. (10) have used ‘H NMR in the determination of the oligosaccharide structures of A active oligosaccharides from hog submaxillary mucin and of H active structures of hog gastric mucin (11). The oligosaccharides isolated from gastric mucin of cystic fibrosis patients (12) are structurally related to the oligosaccharides isolated from the LN cyst glycoprotein, which have been studied by high field proton NMR as well as NMR (13, 14). In all the above studies, analogies between the chemical shifts of certain “structural reporter group resonances” were used to make proton reso- nance assignments as well as structural assignments of some of the oligosaccharides. However, it is known that under certain conditions, it is possible to utilize nuclear Overhauser enhancement (NOE’) between anomeric protons and the agiy- conic protons to establish the linkage of residues (15, 16). If one can correlate the anomeric and aglyconic protons of the individual residues by means of spin decoupling, the entire structure of an oligosaccharide can, in principle, be deter- mined by NMR experiments only without resorting to ancil- lary techniques such as enzyme degradation, methylation analysis, and mass spectrometry. In spite of some ~ f ~ c u l t i e s with this method, principally resulting from very small NOE expected for oligosaccharides whose rotational correlation times approximately equal the reciprocal of the NMR spec- trometer frequency, we have found it to be very effective in a series of oligosaccharides of 6-8 residues in length isolated from human milk (17).

We have recently used HPLC to isolate all the oiigosaccha- rides from two different ovarian cyst glycoproteins (13, 18). In this paper we report the isolation from an ovarian cyst mucin of A active oligosaccharides by HPLC and the appli-

‘The abbreviations used are: NOE, nuclear Overhauser effect; HPLC, high pressure liquid chromatography; DSS, 4,4-dimethyl-4- silapentane-l-sulfonate.

1599

1600 ' H NMR of Blood Group A Oligosaccharides

cation of NMR including complete proton assignments and NOE experiments to determine the carbohydrate structures.

EXPERIMENTAL PROCEDURES

Materials

The glycoprotein designated "RG" was prepared from pseudomu- cinous fluid of a cystic ovarian carcinoma (patient blood group A) by methods previously described (19). The A activity of the glycoprotein was determined by a hemagglutination inhibition assay. Commercial anti-A (Ortho Diagnostics, Raritan, NJ) was diluted in phosphate- buffered salinetto give 2+ agglutination of a 2% suspension of group A1 red blood cells. Doubling dilutions of the RG ovarian cyst glyco- protein were prepared in phosphate-buffered saline and the amount of glycoprotein determined that completely inhibited hemagglutina- tion was determined. The amount was found to be 0.25 pg in a total volume of 0.3 ml.

Methods

HPLC-Isolation of oligosaccharides from the glycoprotein fol- lowed the procedure of Iyer and Carlson (20). 1.0 g of cyst glycoprotein was dissolved in 100 ml of 1.0 M NaBHl in 0.05 M NaOH and incubated for 16 h in a shaker bath at 50 'C. The reaction was stopped by cooling and neutralization to pH 7.0 with HCI. After concentrating to dryness in a rotary evaporator, borate was removed by repeated addition and evaporation of absolute methanol. When the dried product was dissolved in distilled water and centrifuged, less than 2% of the starting material was removed as solid. The clear supernatant was fractionated on a Bio-Gel P-6 (200-400 mesh, Bio-Rad) column of 2.0 X 100-cm using water for elution. The fractions were monitored for neutral sugar by the phenol-sulfuric acid test (21).

The HPLC apparatus consisted of an LDC Constametric I11 pump, a Rheodyne model 7125 injector and a Kratos SF 770 variable wavelength ultraviolet detector operated in the 200-215 nm range. An octadecyl silica column (DuPont Zorbax, 250 X 4.6 mm) was eluted with water for reverse phase chromatography while normal phase chromatography utilized a Varian Micropak AX-5 column (300 X 4.0 mm) eluted with acetonitri1e:l.O mM KHzPO, (60:40). Small scale preparative chromatography was carried out on the same col- umns by collecting the fractions manually at the detector outlet. Carbohydrate analysis were done by HPLC of the benzoylated methyl glycosides by the method of Jentoft (22).

NMR-For 'H NMR spectroscopic analysis, the samples were dissolved in D20 and then repeatedly exchanged with D20 at room temperature followed by lyophilization. The samples were dissolved in 0.3 ml of high purity DzO (Merck, Sharp and Dohme Co.) in a 5- mm tube and studied at 300 MHz in a Nicolet spectrometer equipped with 1280 computer and 293-c pulse programmer. Typically 4K to 8K spectral data points were collected over a %KHz spectral width. Two-dimensional spin correlated spectra (COSY) were measured using the pulse sequence 90"-tl-90"-t2(acquire). The data were apod- ized with a sin' function and displayed in magnitude mode. Spin difference decoupling spectra were obtained by subtracting control from irradiated spectra and the difference curves were compared with simulated spectra (23). NOE experiments on hexasaccharides and larger oligomers were done at 5 "C and smaller oligomers were studied at 75 "C to overcome effects of unfavorable rotational correlation times (23). The pulse sequence for NOE experiments utilized a 3-s pre-irradiation followed by a 90" observation pulse and acquisition with the irradiation off. The observed chemical shifts are reported relative to internal sodium 4,4-dimethyl-4-silapentane-l-sulfonate using acetone as an internal standard (6 = 2.225 ppm downfield from 4,4-dimethyl-4-silapentane-l-sulfonate).

Signals of exchangeable amide protons were observed in a 9010 mixture of H20 and D20 with 0.01 M trifloroacetic acid to maintain sufficiently acidic pH to slow amide proton exchange (24). Spectra were observed with a selective excitation of the Redfield 2-1-4-1-2 type (25).

Exoglycosidase Digestion-Exoglycosidase digestions were per- formed with a-N-acetylgalactosaminidase from limpet (26). 10-20 nmol of oligosaccharide were incubated in 100 pl of 0.01 M citrate buffer (pH 4.0) at 37 "C with 0.1 unit of enzyme. Aliquots drawn directly from the incubation mixture were injected in the HPLC column for analysis on the reverse phase column. The products were identified by their HPLC retention times by comparison with the authentic H active oligosaccharides previously isolated from the cyst

glycoprotein LN (13). The concentrations of product and reactant

of HPLC peak areas. were estimated over the time course of the digestion by observation

RESULTS AND DISCUSSION*

All of the 14 oligosaccharide alditols which were pure by HPLC criteria were studied by proton NMR spectroscopy and the structures of 13 of them were unambiguously determined. The 'H chemical shifts of compounds R1, R2, R3, and R4 identify them as GalNAc-01, @-D-Galp(l+3)GalNAc-ol, ~-~-Galp(l+3)(/3-~-G1cNAcp(l+6)~GalNAc-ol, and a-L- Fucp( 1+2)-/3-~-Galp(1+3)-GalNA~-ol, respectively, since the chemical shifts of the structural reporter protons are identical to those of the same compounds isolated previously (12, 13).

The carbohydrate analysis (Table 111) and proton NMR spectrum of compound R6 shows it to be a tetrasaccharide alditol with 2 a-linked and 1 @-linked residues. Comparison of our spectrum with that of an A active tetrasaccharide from porcine submaxillary mucin reported by van Halbeek et al. (10) strongly indicates the structure shown for R6 in Table I. We have also compared our spectrum with that taken under identical experimental conditions for an authentic sample of the porcine submaxillary mucin tetrasaccharide generously provided by Dr. T. Gerken of Case Western Reserve Univer- sity. Although we are in agreement with the NMR spectrum and structure reported by van Halbeek et al. (10) we find it necessary to revise some of their proton assignments, specif- ically the interchange of the Fuc and GalNAc anomeric proton assignments and a modification of the galactose H3 and H4 assignments. Our results, summarized in Table IV were based on the COSY data of Fig. 3 which is supported by one- dimensional decoupling and simulation of all multiplet struc- tures, including those distorted by departure from weak cou- pling. Each subregion of the spectrum was integrated to verify that all the spectral intensity was accounted for by the proton assignment. In addition, the exchangeable amide protons of GalNAc and GalNAc-ol were assigned by decoupling to verify the assignments of their respective H2 signals.

A detailed and precise proton assignment for this tetrasac- charide was necessary for interpretation of the NOE data which are somewhat unusual. Irradiation of the anomeric proton of GalNAc gives, in addition to the intra-ring effect at H2 which is characteristic of a-linked sugars, a very small enhancement at the resonance of Gal H3 but a large effect at H4 adjacent to the position of the linkage. This result con- trasts with the more commonly observed case for oligosaccha- rides in which NOE is strongest at the proton attached to the carbon atom involved in the intersaccharide linkage (16, 27). The data for R6 show that this oligosaccharide adopts a conformation in which the anomeric proton of GalNAc is closer to H4 of galactose than it is to the proton at the linkage position. This same conclusion was reached by Lemieux et al. (28) on the basis of NMR data and conformational energy calculations on synthetic oligosaccharides which model the non-reducing terminal trisaccharide of the blood group A active structures. A complete calculation of all the NOE by the method of Brisson and Carver (29) for a conformation

Portions of this paper (including "Results and Discussions," Figs. 1, 2, 4-11, and Tables 11-V) are presented in miniprint at the end of this paper. Miniprint is easily read with the aid of a standard magnifying glass. Full size photocopies are available from the Journal of Biological Chemistry, 9650 Rockville Pike, Bethesda, MD 20814, Request Document No. 85M-1309, cite the authors, and include a check or money order for $7.60 per set of photocopies. Full size photocopies are also included in the microfilm edition of the Journal that is available from Waverly Press.

R5

R6

R7

R8

R9

' H NMR of Blood Group A Oligosaccharides 160 1

TABLE I Structures of blood group A oligosaccharide alditok from the ovarian cyst mucin

Notation Primary structure" -

R1 GalNAc-ol R2 Gal(pl+3)GalNAc-ol

Gal(fl143)

R3 1

7 GalNAc-ol

G lcNAc(o14) R4 Fuc(al~2)Gal(pl~3)GalNAc-ol

Gal(B1-3) 1

7 GalNAc-ol

F u c ( a l 4 2 ) I

7 G a l ( ~ l 4 ) G l c N A c ( f l l + 6 )

GalNAc(al+3) Fuc(a1-2)

1

7 Gal(@l-+3)GalNAc-ol

GalNAc(al-+B) Fuc(a1-2)

1 Gal(P1-3)

7 i

7 GalNAc(al-+3) GalNAc-ol

GlcNAc(B14) Fuc(a142)

Gal(B1-3) 7 1

7 GalNAc(a143) GalNAc-ol

Gal(fl l4)GlcNAc(fll-&)

I

7

Fuc(a1+2)

Gal(~l+3)GlcNAc(/3l~3)GalNA~-ol

GalNAc(a1-3) Fuc(a1-2)

I

7 Gal(~l-3)GlcNAc(fll-3)Gal(Bl-3)

GalNAc(a1-3) 1

GalNAc-ol

R10

R11

R12

GalNAc(a1-3)' Fuc(a1-2) GalNAc-ol

I 7 Gal ( f l l 4 )GlcNAc(~ l -&)

7 GalNAc(a1-3)

Fuc(al-+2) I Gal(fll-3)GlcNAc(~l-+3)Gal(~l-+3)

7 1

Fuc(a1-2) GalNAc(a1-3)

I 7 G a l ( f l l 4 ) G l c N A c ( f l l 4 )

7 GalNAc(a1-3)

GalNAc-ol

1602 'H NMR of Btood Group A Oligosaccharides

TABLE I-Continued Notation Primarv structure'

Fuc(crl-+2)Gal(&+3) I

7 R14 FuC(cui-+z) GalNAc-ol

G a l { ~ l ~ ) G l c N A c ( ~ l ~ ) L

/I GalNAc(al-+3)

All monosaccharide units are D sugars except for Fuc, which is an L sugar.

I

0

I

aR

e.... . .. ._ . .. .. . ,. ..~. .&

0 . . . . . . .. Q

, 8 . ~. I , I , I , , , ..---, r,""" fa 5 5 5 0 9.5 4 0 3 5 w m

FIG. 3. 300 MHz 'H correlated spectrum (COSY) of tetra- saccharide alditol R6 in DBO at 24 "C. The spectral width was 4 700 Hz and the data set consisted of 512 X 512 points in t l and t2 dimensions. The solid line shows the connectivities of the scalar coupled peaks of Gal and the broken line shows those of GalNAc. For Fuc (dotted line) hrther cross-peaks from H2 are not observed owing to the overlap of HZ, H3, H4 resonances at 3.83 ppm.

deduced from preliminary conformational calculations gives results agreeing with this interpretation. In the NOE results on the oligosaccharides from the cyst glycoprotein RG, we found for all the structures in Table I that the major NOE on irradiation of the anomeric proton of GalNAc3 is at Gal H4 implying that the conformation must be similar to that pro- posed by Lemieux et al. (28) for their model trisaccharide.

The NOE results for irradiation of the other anomeric protons of R6 were normal. Irradiation at the resonance of p - Gal H1, showed NOE at GalNAc-ol H3 and irradiation at fucose H1 gave NOE at Gal HZ. In R6, the resonances of fucose HZ, H3, and H4 are strongly coupled (see Table IV) leading to a distortion of the anomeric multiplet by virtual coupling. This result contrasts with that for the trisaccharide alditol a-L-Fucpfl~Z)-P-I)-Gal~(l~)-GalNAc-oi, in which the resonance of Fuc H3 is shifted downfield by 0.13 ppm, removing the strong coupling so that the fucose anomeric doublet is undistorted (23).

The 'H NMR spectra of the remaining oligosaccharides in this series have not been previously reported but their struc- tures were determined unambi~ously from the 'W NMR data by the following method. The spectra were fully assigned by

spin difference decoupling spectra and spin simulation of the difference spectra (23). Therefore all chemical shifts are de- termined to -+1 Hz except in cases of strong coupling for which data are reported to only 0.01 ppm in Table IY. The vicinal proton coupling constants reported in Table V, which were determined to k0.3 Hz except in cases of very strong coupling, prove the stereochemistry of the pyranosides and hence identify the component monosaccharide residues. Strong coupling between H2, H3, and H4 of fucose occurs in R6 as well as in R7, R8, and R11 all of which have the same chain attached to C3 of GalNAc-ol. H6 and H6' of both Gal and GalNAc are very nearly chemically equivalent preventing determination of J5.6 in most cases. Although the coupling constants for GalNAc-ol were explicitly measured only for R5 and R6, no significant differences in line shapes were observed for the other oligosaccharides. For @-residues with axial H1, intra residue NOE to the syn-axial protons at H3 and H5 served to confirm the assignment and stereochemistry. NOE data between protons of glycosidically linked residues were used to determine sequence of the residues. The positions of the glycosidic linkage were determined from the NOE data and from the chemical shifts of the anomeric proton and of all the protons of the aglycone residue. Completely assigned 'H NMR spectra and NOE data for all the linkage types detected in the RG oligosaccharides, which were available from the data of Ref. 23 and for R6 above, provided detailed analogies for the NOE data and for the chemical shifts of both the anomeric and aglycone protons.

CONCLUSIONS

The assignment of the anomeric proton signals of a-Fuc and a-GaiNAc in R6 was made by van Haibeek et aE. (10) on basis of their line shape. While our NMR spectrum and structural assignment agrees with that of van Halbeek et al. (lo), we have revised the assignments of GalNAc H1, H3, and H4 and of Fuc HI on the basis of decoupling to the amide proton of GalNAc and by HOE experiments. The distorted line shape for fucose H1 arises from strong coupling of HZ with H3 and H4 (Table 1%'). The distortion of the H1 line shape by virtual coupling is readily shown by spin simulation (31, 321. Explicit assignment of the fucosyl resonances by decoupiing and NOE shows that their chemical shifts are very sensitive to the identity and linkage to the aglycone (23).

The effective use of NOE to establish connectivity of resi- dues in oligosaccharides is complicated by the dependence of NOE on rotational correlation time. Since small molecules have positive NOE, and large ones have negative NOE it is clear that for some intermediate sized molecules (those for which the rotational correlation time corresponds to the re- ciprocal of the spectrometer frequency) the effect must be zero. For reduced oligosaccharides of the type we have isolated from mucins, it appears that pentasaccharides often meet these criteria in a 30&MHz spectrometer. However, we have observed that the rotational correlation time depends strongly on temperature giving the experimentalist some control over

’W NMR of Blood Group A Oligosaccharides 1603

the sign of the NOE. In experiments on hexasaccharides R5 and R8 at 5 “C, modest negative effects were detected and use of low temperature on all the larger oligosaccharides enhanced the negative effect, Positive NOES were observed for tetra- (R6) and pentasaccharides (R7 and R9) at a temperature of 75 “C. Although the maximum effects are small, they are reliably measurable with sample sizes of 2-5 irmol and ade- quate NMR observation time. This dependence of rotational correlation time on temperature results not from temperature dependence of the oiigosaccharide conformation but rather from the temperature dependence of the extensive water binding of oligosaccharides (23).

In recent years, many research groups have used proton NMR spectroscopy to assist in structural assignments of complex oligosaccharides isolated from various types of gly- coproteins. The principal method used is chemical shift anal- ogies among structural reporter groups, mainly anomeric pro- tons, as introduced by Vliegenthart and co-workers (11, 12). In the absence of assigned NMR spectra of sufficiently similar oligosaccharide structures, the chemical shift analogies are impossible and other methods such as methylation analysis and enzymatic degradation become necessary. If complete proton assignments can be carried out, then the chemical shifts of the aglycone protons can be used to extend and improve the chemical shift analogies giving stronger proof of structure (15). The use of inter residue NOE in combination with a complete proton a s s i ~ m e n t has been proposed as an absolute method for structure assignment (27) on basis of the assumption that the major NOE on irradiation of the ano- meric proton would be at the proton attached to the aglycone position. Koerner et al. (33) have extended this proposal, claiming that the appearance of NOE at the aglyconic proton unambiguously shows the positions of both a- and @-linkages. Bock et al. (34) have invoked the “exo-anomeric effect” to explain the generality of the observation that the proton linked to the aglycone carbon is close to the anomeric proton to which it is linked. However, it is clear from our results that complex oligosaccharides may adopt conformations in which the aglycone proton is not the one nearest to the anomeric proton. Therefore, some caution must be exercised in the use of proton NOE to directly infer the positions of intersacchar- ide linkages in cases for which the conformation of the linkage is not fully understood.

Comparison of the reverse phase HPLC retention times for the oligosaccharide isolated from the A active cyst glycopro- tein RG with those from glycoprotein LN (13) shows that addition of a residue of a-GalNAc to a fucosylated @-galactose residue reduces the retention time. The retention times of the pairs of oligosaccharide alditols which differ by 1 a-GalNAc residue show that the homolog with the added GalNAc has less column retention and therefore less exposed hydrophobic surface. Nevertheless, the retention times of the RG oligosac- charides were adequate for their separation by reverse phase HPLC. Only the mono- and disaccharides were not well separated and required normal phase HPLC for isolation. As in the case of the reduced oligosaccharides isolated from the cyst glycoprotein, LN, we observed that the structure 0-1,-

Fucp( l”r2)@-~-Galp( 1-+3)GalNAc-ol gives the most reten- tion.

Of the oligosaccharide structures shown in Table I, R1, R2, R3, and R4 have been isolated from numerous mucin sources and the tetrasaccharide, R6, has been isolated from porcine submaxillary mucin (10). R8 has been isolated from human gastric mucin (9) but the remaining structures have not been previously reported. All the larger oligosaccharides have at their nonreducing terminals the branched structure WD-

GalNAcp(l4)(cu-L-Fucp(l-+2))-P-galactose and most have two such structures consistent with the very high blood group A activity shown by the glycoprotein. Most of the structures isolated from the A active cyst material, RG, are quite anal- ogous to the LN oligosaccharides with the addition of a residue of a-GalNAc to the fucosylated galactose. R10 and R12 are also based on this core structure but with a type I extension on the 3 arm. Although the core structures of all the oligosac- charides isolated in this study are included among those isolated from the products of Smith degradation of an A active cyst glycoprotein (MSS), Kabat and co-workers (8) found additional structures with further elaboration of both the 3 and 6 arms as well as structures with branching at @-galactose. Either the glycoprotein MSS contains a wider variety of oligosaccharide core structrues than does RG or we have failed to identify certain larger core structures of RG perhaps be- cause they appear in very small quantity in the glycopeptide material of pool IV (Fig. 1).

All of the oligosaccharides isolated in this study had at the reducing terminal GalNAc-01. Since the products of a “peel- ing” reaction would be expected to be oligosaccharides of a modest size, our failure to isolate them from pools 1-111 of the Bio-Gel c~omatogram (Fig. 1) suggests that no base degra- dation of the oligosacch~ides of glycoprotein RG occurred under our reaction conditions. Furthermore, since only a small amount of solid material was discarded in the centrifugation step, the chromatogram of Fig. 1 must represent all the neutral carbohydrate present on glycoprotein RG. Unlike the glyco- proteins LN (13) and EA (181, from which essentially all the neutral carbohydrate was isolated as oligosaccharide, the A active glycoprotein RG gave neutral sugar in the excluded volume of the P-6 chromatogram, and only 75% of the re- covered carbohydrate was identified as oligosaccharide. Ap- proximately half of the remaining 25% of neutral sugar ap- peared by NMR spectroscopy to be a heterogeneous mixture of glycopeptides presumably resulting from incomplete deg- radation of the glycoprotein. Since we were unable to study the oligosaccharide structures in pool IV of Fig. 1, it is possible that they differ from the structure reported in Table I. The material in the excluded peak (pool of Fig. 1) appears not to be a product of incomplete degradation but rather to be a polysaccharide with a small repeating subunit having blood group A active side chains. Its NMR spectrum shows fu- cose(al-+Z), GalNAc(a1-+3) linked to P-galactose in a type I linkage attached to a polysaccharide backbone of @-residues whose linkage and composition are currently under study.

A c k ~ w l e d g ~ n ~ s - W e thank Drs. Y. T. and S. C. Li of Tulane University for providing the ~-~-acetyl~alactosaminidase and Dr. Thomas Gerken of Case Western Reserve University for the sample of porcine submaxillary mucin neutral tetrasaccharide. Albert Lee carried out the carbohydrate analyses.

1. 2.

3.

4.

5.

6.

7.

8.

REFERENCES Kabat, E. A. (1982) Am. J. Ctin. Pathol. 78, 281-292 Rovis, L., Anderson, B., Kabat, E. A., Gruezo, F., and Liao, J.

Rovis, L., Anderson, B., Kabat, E. A., Gmezo, F., and Liao, J.

Newman, W., and Kabat, E. A. (1976) Arch. Biochem. Biophys.

Maisonrouge-M~Auli€fe, F., and Kabat, E. A. (1976) Arch.

Lundblad, A., and Kabat, E. A. (1971) J. Immunol. 106, 1572-

Wu, A. M., Kabat, E. A., Pereira, M. E. A., Gruezo, F. G., and

Wu, A. M., Kabat, E. A., Nilsson, B., Zopf, D. A., Gmezo, F. G.,

(1973) Biochemistry 12, 1955-1961

(1973) Biochemistry 12,5340-5354

172,535-550

Biochem. Biophys. 175,90-113

1577

Liao, J. (1982) Arch. Biochem. Biophys. 215 , 390-404

and Liao, J. (1984) J. BioL Chem 259, 7178-7186

1604 ‘H NMR of Blood Group A Oligosaccharides

9. Slomiany, A., Zdebska, E., and Slomiany, B. L. (1984) J. Biol. Chem. 259 , 14743-14749

10. van Halbeek, H., Dorland, L., Haverkamp, J., Veldink G., Vlie- genthart, J. F. G., Fournet, B., Ricart, G., Montreiul, J., Gath- mann, W., and Aminoff, D. (1981) Eur. J. Biochem. 118,487- 495

11. van Halbeek, H., Dorland, L., Vliegenthart, J. F. G., Kochetkov, N. K., Arbatsky, N. P., and Derevitskaya, V. A. (1982) Eur. J. Biochem. 127.21-29

12. van Halbeek, H., Dorland, L., Vliegenthart, J. F. G., Hull, W. E., Lamblin, G., Lhermitte, M., Boersma, A,, and Roussel, P. (1982) Eur. J. Biochem. 127, 7-20

13. Dua, V. K., Dube, V. E., and Bush, C. A. (1984) Biochim. Biophys.

14. Bush, C. A., Panitch, M. M., Dua, V. K., and Rohr, T. E. (1985) Anal. Biochem. 145,124-136

15. Dabrowski, J., Hanfland, P., Egge, H., and Dabrowski, U. (1981) Arch. Biochem. Bwphys. 210,405-411

16. Brisson, J.-R., and Carver, J. P. (1983) J. Biol. Chem. 258,1431- 1434

17. Dua, V. K., Gosso, K., Dube, V. E., and Bush, C. A. (1985) J. Chromatngr. 328, 259-269

18. Dua, 1 7 . K., Dube, V. E., Li, Y.-T., and Bush, C. A. (1985) Glycoconjugate J. 2, 17-30

19. Kabat, E. A. (1956) Blood Group Substances, Their Chemistry and Immunochemistry, pp. 125-129, Academic Press, New York

Acta 802 , 29-40

Supplementary material t o

CHARACTERIZATION OF THE OLIGOSACCHARIDE ALDITOLS FROM OVARIAN CYST MUCIN GLYCOPROTEINS OF BLOOD GROUP A

USING HPLC AND HIGH FIELD ‘H NMR SPECTROSCOPY

Virendra K. Dua, E. N. Narasinga Rao, Shing-Shing Wu, Volker E. Dube and C. Allen Bush

Results and Discussions

A typical elution curve for the blood group A ovarian cyst oligosaccharide alditols on the Bic-Gel P-6 column is given in Fig. 1. The fractions were pooled as indicated.

0

t

E, m

Fraction Number

Fig. 1. Bio-Gel P-6 fractionation of the products of alkaline borohydride degradation of ovarian cyst glycoprotein RG. Detection of neutral carbohydrate by phenol-sulfuric acid test. Fraction volume was 5.1 ml.

20.

21.

22. 23.

24.

25. 26. 27.

28.

29.

30. 31.

32.

33.

34.

Iyer, R. N., and Carlson, D. M. (1971) Arch. Biochem. Biophys.

Dubois, M., Gilles, K. A., Hamilton, J. K., Rebers, P. A., and Smith, F. (1956) Anal. Biochem. 28, 350-356

Jentoft, N. (1985) Anal. Biochem. 148, 424-433 Rao, B. N. N., Dua, V. K., and Bush, C. A. (1985) Biopolymers,

Bush, C. A., Duben, A. J., and Ralapati, S. (1980) Biochemistry

Redfield, A. G. (1978) Methods Entymol. 49,253-270 Li, Y.-T., and Li, S.-C. (1970) J. Biol. Chem. 245, 5143-5160 Koerner, T. A. W., Prestegard, J. H., Demou, P. C., and Yu, R.

Lemieux, R. U., Bock, K., Delbaere, L. T. J., Koto, S., and Rao,

Brisson, J . R., and Carver, J. P. (1983) Biochemistry 22, 1362-

Dua, V. K., and Bush, C. A. (1983) Anal. Biochem. 133,l-8 Perkins, S. J., Johnson, L. N., Phillips, D. C., and Dwek, R. A.

Brisson, J.-R., and Carver, J. P. (1982) J. Bwl. Chem. 257,

Koerner, T. A. W., Scarsdale, J. N., Prestegard, J . H., and Yu, R. K. (1984) J. Carbohydr. Chem. 3,565-580

Bock, K., Meldal, M., D. R., Iversen, T., Pinto, B. M., Garegg, P. J., Kvanstrom, I., Norberg, T., Lindberg, A. A., and Svenson, S. B. (1984) Carbohydr. Res. 130,35-53

142, 101-105

in press

19,501-504

K. (1983) Biochemistry 22,2687-2690

V. S. (1980) Can. J. Chem. 58, 631-652

1368

(1977) Carbohydr. Res. 59,19-34

11207-11209

Since pool V appears at the void volume, it presumably contains material of molecular weight greater than 60M) daltons. The ‘H NMR spectrum of the pooled material of fraction V shows less than 5% amino acid and the rather simple NMR spectrum appears to be that of a polysaccharide with a small repeating unit. The NMR spectrum of pool IV indicates that it is a heterogeneous mixture of glycopeptides and it was not studied further. Pools I to I11 contain 75 $6 of the total glycoprotein carbohydrate and our NMR analysis indicates that the corresponding fractions 47 through 72 contain oligosaccharide alditols ranging in size from one to ten residues.

The reverse phase HPLC chromatogram of the combined pools I to 111 (Fig. 2) shows 14 major peaks each of which was prepared in pure form and structurally identified as will be described below. (See Tab. I for oligosaccharide structures.)

R12

J Gi L

t J , . ”~L”l” d“-1 I. ! I 1 i , i 1 3 5 7 9 11 13 15 17 19 21 22

T I M E I N M I N U T E S

Fig. 2. HPLC of the combined pools I to III from the Bio-Gel P-6 chromatogram of Fig.

at 202 nm. Peaks identified correspond to oligosaccharide alditols whose structures are 1. The column was Zorhax reverse phase, eluted with water a t 1 ml/min and uv detection

given in Tab. I.

' H N M R of Blood Group A Oligosaccharides

Table IL APLC Relative retention value (k') of some oligosxcharide alditols isolated from Blood Group A ovarian cyst Glycoproteins.

Table Tv: 300 MHz. 'H NMR chemical shifts' of the blood group A oligosaccharide

with respect to DSS (indirectly w.r.t. acetone, 6 = 2.225 ppm). alditols of the ovarian cyst mucin in DIO at 24 'C. Chemical shifts are referenced

Compound Relative (Notation from Table I) Retention

R1 1.4 R2 1.5 R3 1.7 R4 3.6 R5 2.0 R6 2.3 R? 2.6

RE 3.1 R9 3.4 R10 4.0 R11 4.7 R12 6.6 R13 7.7 R14 9.5

Chromatographic conditions

column Zorbax ODS solvent water flow rate Iml/min.

Table I11 Carbohydrate analysis of the 'RG' ovarian cyst oligosaccharides.

Oligosaccharide Residuesa

GalNAc GlcNAc Fuc Gal

R5 R6 R7 R8

R10 R9

R11 R12 R14

0.9 1.1

1.0 1.0 2.2

1.2 0.0 1.0 1.2

0.9 1.2 1.0 1.0 1.1 1.0 1.9

1.1 0.9 1.0 0.9

2.3 1.1 2.4 1.0 3.1

2.4 0.7 2.0 2.3 2.4 2.0 3.3

1.2 1.0 2.0 2.3

and taking molar response factors from a standard mixture. GalNAc-ol is a Molar ratios assuming the smallest integral number of fucose residues

not detected by this method (22).

Residue HI 82 H3 HI H5 H61,H62 CHI N-Acetyl

R5 Huuaccharide Fu2' 5.348 5.810 Gal' 4.462 3.573 Gal' 4.594 3.903 GlcNhc' 4.533 3.759 GdNAc' 6.175 4.230 GdNAr-ol 3.770 4.392

R6 Pcntuaccharide h2 5.389 5.83 Gd" GdNAcS 6.189 4.247

4.710 3.904

GalNAs-01 3.805 4.304

R7 Pentuaccharide Fucl G d S 4.700 3.890

5.381 3.80

GlcNAc' 4.500 3.725 GalNAc' 5.189 4.247 GdNAc-01 3.785 4.309

R8 Bexuaccharidc mc' 5.379 3.80 0 . 1 s Gd'

4.700 3.895

GkNAc' 4.687 3.780 4.472 3.642

GdNA? 6.189 4.249 GdNAc-ol 3.800 4.302

R9 pentaaxcharide (5p. I) Fu? Gal".' 4.714

5.264

GlcNAc' 4.581 GdNAc' 5.187 GalNAs-ol 3.756

R10 Octasaccharide F"2 5.251

Gal' 4 433

Gal'.' 4.704 4.466

GlcNAc' 4.621 GlcNAc' 4.558 GalNAc' 5.181 GalNAc-ol 3.7-3.8

R11 Octaaaccharide Fuc2,9 5.312 Fuel.' 5.348 Gal3 Gal'

4.700

GlcNAc' 4.571 4.597

GzlN~c',~ 5.188 GalNac',' 5.116 GdNAc-ol 3.785

R12 Decasaccharide Fuc'.' 5.252 Fuel.' 5.353

Gal'

G

4.436

4.710 4.598

G L N A ~ ~ 4.599 G ~ N A C ~ 4.524

GalNAc3,' 5.177 GalNAc3,' 5.111

GalNAc-ol 3.800

R14 Heptasaccharide FUC'.' 5 217 FUc=.4 5.349 Gal3 Gal' 4.598 G~NAC' 4.558

4.575

GalNAc3 5.171 GalNAc-ol 3.786

3.773

3.798 3.825

4.280 4.222

3.159 3.554 3 558 3 no6 3.814 3.14 4.233 4.400

3.800 3.795

3.898 3.891

3.779 4.247 4.247 4.270

3.797 3.769

3.554 3.905 3 812 3 825 3.76 4.219 4.237 4.402

3.801 3.794 3.672 3.899 3.774 4.237 4.402

3.659

3.977 3.667

3.70 3.wO 4.077

3.83 4.019 3.930 4.100

3.80 4.011 3.678 3.932 4.081

4.025 3.00

3.670 3.73 3.931 4.095

3.610 3.945 4.037 3.935 8.981

3 605 3.722 3.675 3.940 4.011 3.72 3.951 4.059

3.8W

4.010 3.661

3.910 3.780 3.936 3.910 4.088

3.805 3.654 3.720 3.980 3.935 4.010 3.15 3.949 3.935 4.040

3.920 3.664 3.847 3.984 3.78

4.083 3.904

3.835

4.226 3.895

4.010 3.74

3.468

3.81 4.225 4.019 3.805

3.79 4.215 3.47 4.02 3.695

3.80 4.225 3.920 3.72 4.02 3.590

3.785 4.225 3.51 5.980 3.580

3.805 4.142 3.915 4.223 3.51 3.71 3.967 3.463

3.805

4.220 3.815

4.220 3.80 4.010 3990 3.583

3.782

4.118 3.765

4.213 4.213 3.51 3.73 3.973 4.002 3.445

3.834 3.834 3.930 4.218 3.19 4.010 3.500

4.319 3.7001 3.7001 3.461 4.210 4.270

4.337

4.156 3.717"

4.125

4.336 3.703' 3.47 4.100 4.225

4.325 3.704 3.70. 3.005 4.162 4.220

4.326 3.685" 3.53 4.265 4.116

4.335 3.715' 3.710' 3.660' 3.54

4.295 3.600

4.282

4.322 4.322 3.7101 3.7w 3.565 4.161 4.220 4.220

4.338 4.311 3.727" 3.710' 3.659' 3.53 3.446 4.294 4.221 4.261

4.277 4.317 3 616' 3.675' 3.468 4.235 4.275

1.248

4 . m , 3.810 3.785 3.940, 3.710

1.231

3.770 3.675

1.232

3.942, 3.780 3.788 3.815, 3.704

1.235

4.018, 3.841 3.711 3.920, 3.720

1.243

3.945, 3.76 3.m 3.835

1.245

4.010. 3 195 4.005, 3.803 3.752 3.940, 3.700

1.230 1.248

4 037, 3.160 3.761 3.940,

1.252 1.252

3.903, 4.005.

3.

3.

3. 3.

3.744 1 n n

794

100

794 810

-. .. 3.940, 3.700

1.247 1.247

4.010, 3.810 3.764 3.945, 3.685

2.066 2.046 2.046

2.048 2.048

2.055 2.047

2.049 2.049 2.049

2.117 2.038 2.038

2.063 2.063

2.036 2.063

2.047 2.058

2.047 2.047

2.063 2.063 2 039 2 039 2.063

2.058 2 039 2 058 - -

.I Accuracy of chemical ahifte: i 0.005 ppm. In c-8 where resonances u e strongly coupled they are reported to only two decimal places with an accuracy of i 0.01 ppm. Assignments have been made using ID - difference deeoupling expermint. employing total d u r a t i o n M well M spin-tickling

simulated to verify the observed coupling patterns. techniques. To rnolve ambiguities due to multiple saturation difference-decoupled spectra were

'A suprlacript at the name of a sugar residue indicate. to whlch position of the next monosaccharide it is glycosidically linked (in this case C2 of Gal on the CB-arm of GalNAc-01). Where there is more

for e.g. in R12 - h c 2 3 meam that h c is (1 - 2) linked to a residue (Gal) which Itself ia (1 - 3) than one residue with the same linkage an additional soperscript is uaed for further distinction, M

linked (to GlcNAc). E Assigned by n.0.e.

1606 'H NMR of Blood Group A Oligosaccharides

charides from the 'RG' ovarian cyst. Table:V Observed coupling constants' in the various A-blood group oligosac-

R5 Hexuacshuids

Fuc Gd'

3.7 10.3 8.1 10.0

Gal' GlcNAc 8.0 9.6

7.3 10.3

GalNAs' 3.7 11.4 GalNAc-01 7.3 1.4

R6 Tetruaccharide

Fu2 Gal8

c b 7.4 10.3

GdNAcS 3.8 11.4 GalNAc-ol 7.0 2.2

R7 Pentas..cchuide

Fuc' Gal3 7.3 9.5

c b

GlsNAc' 8.1 10.2 GalNAs' 3.7 11.6

R8 Xexaswchuidc Fu.3 c b Gals G al'

7.7 9.6

GlsNAc' 7.3 10.3 7.7 10.2

R9 Pentasrcchuido Fuc= 4.3 10.3 Gals GIcNAcJ 8.0 10.3

7.3 10.2

GalNAZ 2.9 11.0

R10 Octasacchuide

Fuc= 3.9 10.6 Gal3 7.5 9.8 G 4' 8.2 10.2 Gal'.' 7.8 9.6 GlcNAc' 8.2 10.6 GIcNAsa 8.1 b GaINAs' 3.4 10.3 Rll Octasucharids FUP.' c b Fuc'.' 3.8 10.7 G ala 7.8 9.8 Gal' 7.8 9.7 ~~

GlcNAc' 7.8 10.8 GalNAc'.' 3.8 10.5 GaINAc',' 3.8 10.5

R12 Drasaccbuide Fuc'.' 3.7 10.3 Fuc'.' 3.7 10.2

Gal' Gal'

8.1 9.5 7.4 9.5

GICNAS 8.1 9.6 Gal'J 7.4 9.5

G ~ ~ N A C ' . ~ 3.8 10.9 GlcNAre 7.8 9.6

G ~ I N A C ~ , ~ 3.8 10.0

R14 Octasucharide

Fuc".' 3.7 10.6 Fuca,' 3.6 10.3 Gal3 Gal'

7.5 9.5

G ~ N A ? 7.8 10.2 7.4 10.3

GalNAc' 3.7 11.4

3.3 3.0

3.4 b 3.6 10.4

b 3.5 3.4 10.6

b 3.7 8.6 3.0

b 3.5 3.6 b

3.6

9.5 3.2

3.5

3.0 2.8 3.0 2.9 9.7 b 3.6

b

3.0 2.8

3.0 b 3.5 3.5

3.0 2.9 3.0 3.0 3.0 9.8 b 3.3 3.0

3.0 2.9

3.2 3.0

b 3.0

<1.0 <LO <1.0 9.5

<1.0 1.2

<LO <LO <1.0 1.5

<LO <1.0 b

4.0

<LO <l.O <1.0 9.5

<l.O <LO

c1.0 b

c1.0 c1.0 <1.0 c1.0

b 9.5

c1.0

<LO <1.0 <1.0 <1.0 9.5

<1.0 <1.0

<1.0 <1.0 <LO <LO <1.0

b 9.6

<LO <LO

<1.0 <LO <LO <LO 9.8

<1.0

6.6

5.9

7.0

6.4

5.9 6.7

6.8

5.4

6.3

6.8

6.4

6.5 5.0

6.4

5.5 5.8

6.0 6.0

5.0 6.8

5.8 5.8

5.6 5.8

6.6 6.6

6.0

J W -

2.2

2.0

2.0

2.0

2.1 2.0

2.0

2.1 2.0

2.4

J... -

11.8

11.7

12.0

11.8

12.0 12.0

11.9

12.0 11.8

12.1

Reverse phase HPLC of pool I shows five major peaks, R1, R2, R3, R4 and R6 which were identified as oligosaccharide alditols ranging in size from mono- to tetrasaccharide. Fractions corresponding to peaks R3, R4 and R6 were collected, freeze dried and shown by high field NMR spectroscopy to be pure oligosaccharides. Since the retention times of R1 and R2 in reverse phase HPLC are very similar (Tab. 11). they were collected together

not shown)was similar to that for the same oligosaccharides isolated from glycoprotein and separated by normal phase HPLC as described in ref. 13. The HPLC separation (data

LN (13). Reverse phase HPLC of pool I1 shows eight major peaks R5, R6, R7, R8, R9, R10, R11 and R14 all of which were prepared and shown by high field NMR spectroscopy to be pure oligosaccharides ranging in size from pent& to octasaccharides. Four major peaks (R10 to R13) appear in the reverse phase HPLC of pool 111. These were isolated by pieparative reverse phase HPLC and characterized as octa- to decasaccharides by high field NMR spectroscopy.

in Tab. W . The chemical shifts of all the resonances assigned to Fuc, GalNAc and Gal The 'H NMR spectrum of compound R7 is given in Fig. 4 and the proton assignment

in R7 are similar to those of R6 suggesting that it has the same substitution of Gal3 . The anomeric doublet a t 4.560 ppm and the amide methyl signal a t 2.047 ppm imply an additional residue of acetamido sugar and the coupling constants show that it is 0- GlcNAc. The downfield shift of GalNAc-ol H5 in R1 compared to that in R6 implies substitution by GlcNAc at C6 in addition to Gal a t C3 of GalNAc-ol (13,23).

Fig. 4. 300 MHz 'H NMR apectnun of pentanaceharide R7.

In cases such as R6 where GalNAc-ol is not substituted at the 6 position, the two H6 protons are chemically equivalent. In RI, the substitution of GalNAc-ol by GlcNAc

result of restricted rotation about the GalNAc-ol CS-CB bond. Our data imply that this a t C6 causes the signals of the two GalNAc-ol A6 to become inequivalent apparently as a

inequivalence of the GalNAc-ol H6 protons is a general diagnostic for substitution at C6 of GalNAc-ol (23). The characteristic resonance at 3.47 resulting from strongly coupled signals of GlcNAc H4 and H5 is typical of a GlcNAc residue which is unsubstituted (23). n.0.e. data for R7 show patterns on irradiation of the anomeric protons of GalNAc, Fuc and Gal3 which are similar to those observed for the tetrasaccharide R6. Irradiation of GlcNAc H1 gives n.0.e. at the more upfield of the GalNAc-ol H6 resonances (3.704 ppm) confirming the structure of R7 given in Table I.

Integration of the anomeric region of the spectrum of compound R8 (Fig. 5) shows five anomeric protons which, with the addition of GalNAc-01, implies a hexasaccharide consistent with the carbohydrate analysis (Table 111). The chemical shifts of the charac- teristic resonances of GalNAc-ol are similar to those of R7 implying substitution at C3 by

assigned to a-Fucz H1, a-GalNAc3 H1 and @Gal3 H1 on the basis of their close analogy Gal and at C6 by GlcNAc. The anomeric doublets a t 5.379, 5.189 and 4.700 ppm, were

to the chemical shifts of the same residues in R6 and R7.

"d

i h I # ' l j ;{ 4 y p f l

i ,,

+&:LL+L!i.2*.4-F ?:h+':: :""" 1.8 1.. s.2 s.0 4.a ,e 4.4 4.2 4.0 1.8 1 8 I & 3.2 2.s 2.0 t .6 1.0 r w d

Fa. 5. 300 MIia 'H NMR spectrum of huanaccharide RE.

oligosaccharides implying that R8 has the same type of chain connected to C3 of GalNAc-ol. Similar n.0.e. on irradiation of these anomeric resonances was also observed for all three

R8 must therefore be an extension of R I by the addition of a single residue whose coupling constants indicate it to be P-galactose. Therefore the resonance at 4.472 ppm is assigned to &Gal' H1, and the type I1 linkage to GkNAce is implied by the chemical shifts of Gal' H1 and of GlcNAc H4 and H5. In a typical type I1 chain, such as lacto-N-neotetraose

H4 is shifted downfield into the 3.8 ppm region by galactose substitution leaving GlcNAc (LNneoT), the anomeric proton of Gal' is a t 4.479 ppm (30) while the resonance of GlcNAc

H5 as the lone upfield resonance at 3.55 ppm where it overlaps with GalNAc-ol H4 as shown in Fig. 5, (23). In lacto-N-tetraose (LNT), a type I chain, the chemical shift of the anomeric proton of the galactose which is Pl-3 linked to GlcNAc is 4.436 ppm and GlcNAc H4 and H5 are closely coupled in the 3.45 to 3.55 ppm region (12,23,30).

of GalNAc-01 are similar among R6, R7 and R8. These resonances which are characteristic The chemical shifts of a-GalNAc' and a-Fucz linked to the &Gal on the C3 position

of the blood group A substituents, are found in the apectrum of R11 which also contains this structural feature as will be discussed below. Not only is the signal aasigned to Gal3 H1 strongly downfield shifted but there is also a small upfield shift of the resonances of GalNAc-ol HZ and H5 in comparison to the corresponding blood group H oligosaccharides lacking the a-GalNAc substituent (13,23). In addition to oligosaccharides containing this kind of A active chain, we have also isolated from glycoprotein RG structures having a- GalNAc3 and u-Fuc2 substituents on the P-Gal' residue of type I1 chains attached to the GalNAc-ol C6. The chemical shifts of the structural reporter resonances of such chains in R5, R11, R12 and R14 are distinctive as will be discussed below.

'H NMR of Blood Group A Oligosaccharides 1607

I

I

Fig. 6. 500 MEs 'E N M R spectrum of hexasacchuide RS.

galNAc-ol implying a hexasaccharide. Integration of the amide methyl region shows 9 The NMR spectrum of R5 (Fig. 6) shows five anomeric resonances plus those of

protons of three acetamido groups and decoupling of the amide protons confirms their

5.348 ppm was shown by decoupling to be a-Fuc' and its linkage to galactose is implied by assignment to galNAc-ol, to a-GalNAc3 and to P-GlcNAce . The anomeric doublet at

ppm. The anomeric doublets at 4.462 and 4.549 ppm were shown to be L3 galactme by the chemical shift of its H5 resonance which is coupled to the methyl group doublet at 1.248

their coupling constants and chemical shifts. Theae data, supported hy the carbohydrate analysis, imply R5 to he a hexasaccharide alditol isomeric with R8.

similar to those of the core tetrasaccharide of LN or EA glycoprotein in which galNAc-ol The chemical shifts of the resonances assigned to GalNAc-01 HZ and HS in R5 are

is substituted by gala and glcNAce (13J8.23) and are downfield from those of the isomeric hexasaccharide R8 in which the galactose linked to C3 of GalNAc-ol is substituted by fucose and GalNAc. The chemical shitt of the @-galactose anomeric proton resonance at 4.462 ppm implies that it is not fucosylated and its linkage to C3 of GalNAc-ol is confirmed by the observation of n.0.e. at GalNAc-ol H3 (4.077 ppm) on irradiation of the anomeric proton resonance. That the fucose and GalNAc substituents in R5 are on the type II chain connected to the GICNAC~ was ahown by additional n.0.e. experiments. Irradiation at u- Fuc'" H1 gives n.0.e. at &Gal4 HZ as well as at Fuc HZ and irradiation at a- GalNAc3-' H1 (5.175 ppm) gives n.0.e. at @-Gal' H4. confirming the attachment of both Fuc' and GalNAcS to &Gal' . The chemical shifts of Gal' HZ, H3 and H5 in R5 are close to those of the similarly substituted Gal3 in R6, R7 and R8 and the similar appearance of the n.0.e. observed on irradiation of the anomeric protons of Fuc' and GalNAc' implies that the conformation of their glycosidic bondn in theae two fragments must not differ greatly. The type 11 linkage between @-Gal' HI and GICNAC~ is indicated by the chemical shifts of GlcNAc H3, H4 and HS which are similar to those of the GlcNAc residue in the type

fragment(23). It will he seen below that the chemical shifts of a GlcNAc H3, H4 and H5 II chain of H active oligosaccharides having the Fuc ap(1 -+ 2)Gal /? p(1 -+ 4)GlcNAc

in type I chains differ substantially from those of R5.

suggests six glycosidically linked residues and an alditol expested for a heptasaccharide Both the carbohydrate analysis and the integration of the spectrum of R14 (Fig. 7)

alditol. The chemical shifta of the characteristic resonances of GalNAc-ol of R14 are similar to those of R5 implying that the Gal* linked to GalNAc-01 does not carry both fucose and GalNAc substituents. The amide methyl group. resonances at 2.058 and at 2.039 ppm integrate as six and three protons each and the corresponding amide protons are assigned to @-GlcNAce , GalNAc-01 and a-GalNAc'~' by decoupling to the respective

were identified an fucose. The chemical shifts of their H5 resonances are characteristic of HZ resonances. Integration of the 6denxy hexose methyl region shows huo residues which

galactose by the vicinal coupling constants. The chemical shifts of the resonances assigned a-L-Fucp(1 + 2)Gal linkages. The doubleb at 4.575 and 4.598 ppm were identified as

residues are attached to the type 11 chain which includes ,9-GlcNAce . These linkages are to Fuc'.' , GalNAca , Gal' and GIcNAc" are similar to those of R5 suggesting that these

presence of the a-LFucp(1 -+ l)-p-D.Galp(l "t 3)-GalNAc-ol arm for the structure R14 confirmed by observation of the same n.0.e. relations that were observed in RS. The

shown in Tab. I is suggested by chemical shift analogies to the resonances found for similar H active ntmcturea (23) and the structure was confirmed by observation of n.0.e. at p-

n.0.e. at GalNAc-01 H3 (4.083 ppm) on irradiation of @-GalS H1 (4.571 ppm). Gala HZ (3.672 ppm) on irradiation of ~ - F U C * ~ ~ H1 (5.217 ppm) and by observation of

indicate an octasaccharide alditol. The chemical shifts of the signals assigned to GalNAc-ol The 'H NMR spectrum of olisosaccharide R11 (Fig. 8) and the carbohydrate analysis

HZ and H5 imply substitution at C6 by GlcNAc and at C3 by P-Gal which bears the sub- stituents a-Fuc' and u-GalNAca . Integration of the deoxyhexose methyl region indicates two residues which are related by d&oupling to the two a-fucme anomeric resonances at 5.372 and 5.348 ppm. Integration of the amide methyl region s h m four residues and the amide protons were assigned hy decoupling to two residues of a-GalNAc and single residues

both are linked to Gal C2. The assignment of resonances at 5.348, 5.176, 4.597 and 4.571 of GalNAc-ol and p-GlcNAc. The chemical shifts of the fucose HS resonances imply that

ppm to u-Fuc'.' H1, a-GalNAc3.' H1, 0- Gal' H1 and P-GlcNAce H1 by analogy with the assignments in R5 and R14 suggests a type II chain attached to the C6 position of GalNAc-ol bearlng fucose and GalNAc substituents. These assignments are supported by the same type of decoupling and n.0.e. data which were o b a w e d for R5 and R14. Sup port for the structure of the chain attached to C3 of GalNAc-ol BS indicated in Tab. I is obtained by n.0.e. data to establish the relationship of resonances assigned to m-Fuc2~' H1 (5.372 ppm), u-GalNAcS** H1 (5.188 ppm), &Gal' H1 (4.700 ppm) and GalNAc-ol H3 (4.088 ppm).

containing type I chains Substituted by fucose and GalNAc from the glycoprotein RG. In addition to the StNctUfS described above, we have identified oligosaccharides

In the spectrum of oligosaccharide R9 shown in Fig. 9, the methyl doublet arises from a single residue of fucose and the resonance assigned to H5 at 4.326 implies that it is linked to Gal C2. Integration of the amide methyl region implies three residues which were identified as GalNAc-01, 8-GlcNAc and u-GalNAc by amide proton decoupling. In addition to the 8-anomeric doublet at 4.581 ppm which is assigned to GlcNAc, a second B-anomeric doublet at 4.714 ppm is assigned to galactose. The chemical shifts of the raonanca of GalNAc-ol are characteristic of Substitution by a single residue of GlcNAc

2.117 ppm is characteristic of a GlcNAc residue linked tu the 3 poaition of GalNAc-01. at the 3 position (12,13). Likewise, the downfield shift of the amide methyl resonance at

This linkage of GlcNAc to GalNAc-ol is confirmed hy the observation of n.0.e. at the resonance assigned to H3 of GalNAc-ol on irradiation of the GIcNAc anomeric resonance. The downfield shift of GlcNAc H3 ( 4.037 ppm) along with the strongly coupled H4 and H5 signals at 3.51 and 3.53 ppm is Characteristic of substitution of GIcNAc by Gal' in a type I chain, (23). The type I chain is also confirmed by the observation of n.0.e. at GlcNAc H3 on irradiation of Gala H1. The downfield position of the signal assigned to the anomeric proton of galactose implies its substitution by GalNAc and fucose which is confirmed by the observation of n.0.e. at Gal3 HZ on irradiation of the fucmyl anomeric proton and at Gal3 HI on irradiation of GalNAc H1.

N d

The spectrum of R12 (Fig. 10) with nine anomeric resonances plus those of galNAc-ol is that of a decasaccharide. The carbohydrate analysis combined with the integration of

be galNAc-ol. Two are assigned to a-GalNAc and two are assigned to residues of p- the amide methyl region shows that there are five acetamido sugars, of which one must

GlcNAc. The 5deoxy hexose methyl region indicates two residues, the chemical shifts and coupling constants of which indicate that they are both u Fucp(1 -+ 2)Gal. Of the three anomeric resonances of &galactose, that of Gal' overlaps with the anomeric resonance of GICNAC~ only in the room temperature spectra, not at higher or lower temperatures. The resonances assigned to GalNAc-ol, especially HZ (4.402) and H5 (4.261) ppm, are characteristic of substitution at C6 by GlcNAc and at C3 by a galactose residue which is not substituted by n-GalNAc. Thus R12 is a decuaccharide which we will show to have both type I and type I1 chains substituted by a-fucose and a-GalNAc.

1608 'H N ~ R of Blood Group A Oligosaccharides

in chemical shift8 of a-Fuca*' , a-GalNA&' , @-GkNAca and P-Gal' to those in structures The assignment of a type II chain l iked to GalNAc-01 C6 is suggested by similarities

R5, R11 and R14. This conclusion is supported by n.0.e. data for these resonances which are exactly similar to those reported for R5, R11 and R14. That a type I chain is attached to GalNAc-ol C3 may be established by arguments based on n.0.e. and chemical shift data. First, the linkage of Gal3 (-01) is identified by n.0.e. observed at GalNAc-ol H3

this residue implies that is substituted at the C3 pasition by GlcNAc (30). The chemical (4.040 ppm) on irradiation of H1 (4.436 ppm). The chemical shift of H4 (4.118 ppm) of

shifta of the fucosyl, GalNAc and Gal residues on the type I chain in R12 differ both from those of the type I1 chain (cf. R5, R14) and from those of the same residues which are directly connected to the Gal3 (-01) (cf. R6, R7 and R8) but are similar to those of

similar for all three types of chain. Irradiation at a - F u ~ ~ 5 ~ HI gives n.0.e. at the resonance the type I chain in RQ. In addition, the linkages were confirmed by n.0.e. data which are

assigned to 8-Ga13*3 H2. The type I linbge of Gal' was shown by the downfield shift of GIcNAc' H3 (4.010 ppm) caused by Gal' substitution and by the observation of n.0.e. at 6he resonance of GIcNAc~ H3 on irradiation of GalS H1 (4.710 ppm).

signals at chemical shift8 similar to those of R12 indicating substitution at C3 by galactose The 'H NMR spectrum of the octasaccharide alditol R10 (Fig. 11) exhibib GalNAc-ol

and at C6 by GlcNAc. The spectrum of R10 also shows the narrow multiplet at 4.142

resonances at chemical shifts which we assigned to the residues of the type I chain with ppm assigned to H4 of a galactose residue substituted by GlcNAc at C3. The presence of

the presence of that chain as indicated in the structure shown in Tah. I. This structural ~ - t F ~ c p ( l "t 2)and o-D-GalNAcp(1 -+ 3)substituenta linked to C3 of GalNAc-ol implies

aseignment is confirmed by n.0.e. observed at GalsJ H2 and H4 on irradiation of Fuc

/ / iI

Fig. 11. 300 MHz '€I NMR spectrum of octasacchuide R10.

H1 and GalNAc H1 respectively. The type I chain is confirmed by observation of n.0.e at GlcNAc' H3 on irradiation of Gal',' HI and by u.0.e. at Gal3 (-01) H3 on irradiation of GIcNAc' H1. Signals characteristic of Fnc and OalNAc residues on a type II chain are absent. The assignment of a type I1 chain attached to GlcNAcB is based on analogies of the chemical shifts to those of the similar chain in R8.

aiditol having the same carbohydrate composition as R12. We suspect that the structure The IH NMR spectrum of compound R13 (data not shown) indicates a decasaccharide

of R13 features a type I chain extension of the 6 arm but the quantities of this minor constituent which we have isolated are too small to allow confirmation of this speculation by n.0.e. data.

Enzymatic digestion of several of the oligosaccharides by exoglycosidase provided fur- ther evidence for the structures given in Table I. Digestion of the oligosaccharides of Table

which have been isolated previously from the ovarian cyst glycoprotein, LN. They can be I by a-N-Acetyl galactosaminidase is expected to lead to oligosaccharideproducta, many of

identified with reasonable certainty since their retention times in reverse phase HPLC have been shown to vary widely (13). Digestion of oligosaccharide R6 for 16 hr at 37 "C resulted

of the expected trisaccharide, R4. A 16 hr incubation of oligosaccharide R14 resulted in in complete conver+n to a product whose retention time corresponded exactly with that

complete conversion to the expected ~ f u c o h e x ~ ~ c h a r i d e which was called L10 in ref. 13. For R11. which has two o-GalNAc residues, a 16 hr. digestion resulted in complete loss of the substrate and appearance of a mixture of R14 and L10 in a 1 to 3 ratio while further incubation gave L10 as the sole product. Incubation of the type 1 pentasaccharide, RQ for

L7 in ref. 13. 16 hr. resulted in complete conversion to the expected tetrasaccharide which was called