their oligosaccharide structure or that the N antigen is biosynthetically ...... ganglioside^,^ or Forssman hapten (26) and are as follows (+. 0.2 ppm): Galal-3 (96.9) ...
THEJ O U R N A L O F BIOLOGICAL CHEMISTRY Vol. 256. No. 11, Iswe of dune 10. pp. 5781-5791. Prmted
In
19H1
(/.SA
Chemical and Carbon-13 Nuclear Magnetic Resonance Studies of the Blood Group M and N Active Sialoglycopeptidesfrom Human Glycophorin A* (Received for publication, October 30, 1980, and in revised form, February 13, 1981)
Rainer Prohaska$g, Theodore A. W. Koerner, Jr.$$,Ian M. Armitage7 11 **, and Heinz Furthmayrg ** Q§ From the Departments of 0 Pathology, 7 Molecular Biophysics and Biochemistry, and #Laboratory Medicine, Yale University School of Medicine, New Haven, Connecticut 06510
N antigen is The NH2-terminal sialoglycopeptides from human their oligosaccharide structure or that the erythrocyte glycophorin A have been obtained by spe- biosynthetically transformed to the M antigen by siacific proteolytic cleavage and gel filtration chromatog- lylation. raphy. By cyanogen bromide cleavage, a glycosylated (*) octapeptide was obtained from blood group M donors having an amino acid composition and 13C NMR Although the primary structureof the glycophorin A polyspectrum consistent with the structure peptide has been elucidated (l),many aspectsof the structure Ser.S&-dr=dr.Gly*Val-Ala-Hse (1) of this complex cell surface glycoprotein remain obscure. One reason for thisis the fact that15 oligosaccharide units are 0Similarly, from blood group N donors, the structure glycosidically linked to the polypeptide chain at certain serine Leu.S~r=T$r*T$r*Glu*Val.Ala.Hse (11) and threonine positions within the first 50 residues of the was demonstrated. By Staphylococcus aureus protease amino acid sequence and there is little informationavailable cleavage, a glycosylated pentapeptide was obtained on the individual oligosaccharide structures at each attachfrom N donors having the same structure as 11,without ment site. Moreover, virtually nothing is known of the effect of these oligosaccharide groups on the conformation of the the carboxyl-terminal sequence Val Ala Hse. of the 0Methanolysis/gas chromatographic analysis and 13C polypeptide chain. Previous studies on the structure NMR spectroscopy of I and 11 and their asialo deriva- linked oligosaccharides have been based on material prepared tives reveal that the M- and N-active sialoglyco-octa- from mixtures of sialoglycoproteins of red cell membranes by peptides bothhave identical oligosaccharide struc- alkalineborohydridetreatment.Thedatathusobtained tures, each containing three 0-linked tetrasaccharides have suggested the presence of tetrasaccharides having the NeuNAca2-3Ga1/31-3(NeuNAca2- structure NeuNAccu2-3Gal/?l-3(NeuNAccu2-6)GalNAcl-Owith thestructure G)GaINAcal-O-Ser(Thr). The demonstration of the an- Ser(Thr)’in addition to trisaccharides with only one NeuNAc omeric form of GalNAc-peptide linkages revealed by residue, but alternative structures have been reported (reI3C NMR has previously been inaccessible by chemical viewed in Refs. 2, 3). analysis. Conformationally, I and 11 appear identical GlycophorinA occursin two formswith differencesin and both manifest several unusual resonance shifts amino acid residues in positions one and five of the sequence suggestive of a glycopeptide secondary structure in(4-6). This difference is correlated with the expression of M volving four specific hydrogen bonds. Calcium ion tior N blood group activity (4) and for most antisera this activity tration was also found to induce shifts in the NeuNAc is determined predominantly by the NH2-terminal residue (7). 13C resonances that may be of functional significance. However, it has been known for quite some time that removal Serological studies reveal that both the M and N glyco-octapeptides and the N glyco-pentapeptide retain of sialic acid residues by enzymatic or mild acid treatment by a large number but not Deam- destroys the activity, as measured all of the M or N activity of the parent structure. ination and/or desialylation completely destroys this all antisera (8, 9). From these findings it has been deduced is determined by carbohydrates activity. These data are consistent with a model in that the antigenic sitelargely which the M or N determinant is the NH2-terminal and furthermore, that thetwo antigens, N and M, are related amino acid and a NeuNAc residue(s). From these data with N being the biosynthetic precursorof M (10). According it is concluded that there is no chemical basis for asser- to the scheme, the N-active peptide is converted to M by tions in the literature that M and N antigens differ in addition of a residue of NeuNAc to one or moreoligosaccha* 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. Present address, Department of Biochemistry, University of Vienna, School of Medicine, Vienna, Austria. 11 Supported by United States Public Health Service Grant AM 18778 and National Science Foundation Grant PCM 77-18941. * * TOwhom correspondence should be addressed. 08 Supported by United States Public Health Service Grant GMAM 21714 and the recipient of Faculty Research Award No. 177 from the American Cancer Society.
*
’ Abbreviations used are: NeuNAc, N-acetylneuraminic acid (a sialic acid); Gal, galactose; GalNAc, N-acetylgalactosamine; Glc, glucose; GlcNAc, N-acetylglucosamine; Rha, rhamnose (all sugars are pyranose forms and are D unless labeled L); /3, the /3-anomer of the indicated saccharide; C”, the carbon two bonds removed from the carbonyl carbon of a peptide residue; AO, the glycosylation shift (shielding effect) for an atom two bonds removed from added glycosyl group; A@, chemical shift difference at peptide C” carbon; M and N designate the blood group phenotype of red cells and isolated peptides; CNBr, cyanogen bromide. Superscripted numbers with amino acid abbreviation refer to peptide residue position (e.g. .’Thr, threonine residue in peptide position three).
5781
5782
Structure of the M N Blood Group Antigens
rides. This interpretationwassupportedbydatashowing apparent differences in the carbohydrate structure of M and N. These observations led to the idea that the M a n d N genes code for sialyltransferases (11). In previous studies on small peptides derivedfrom glycophorin of both types, no difference in carbohydrate compositionwas found (6, 12). The chemical nature of the antigenic determinants was, however, not defined and could still be explained by several alternatives (4). We have undertaken this study on the NH2-terminal pentaor octapeptides of glycophorin AM and AN in order to deterof the polypeptide and mine the composition and contribution carbohydrate structures to the antigenic site important for the interaction with specific antibodies. In addition to chemical and serological analyses, we have subjected the NH2-terminal sialo-octapeptides of glycophorin A to a detailed I 3 C NMR analysis. To our knowledge, this is t h e first reported 13C NMR study of sialoglycopeptides and of membrane-derived glycopeptides. In the courseof this study, assignments were made for the carbons of the G a l N A c a l - 0 Ser linkage and for N-linked homoserine that should aid in future studiesof glycopeptides and cyanogen bromide cleaved
second UV-absorbing peak (230 nm) were pooled and lyophilized. Yield: 9.8 mg. Octapeptide (30 mg) was dissolved in 3 ml of25mM sulfuric acid and heated for 1h a t 80 "C. After cooling, the solution was applied to a BioGel P-6 column (1.5 X 145 cm, 50 mM acetic acid) and fractions of the main UV-absorbing peak (230 nm) were pooledand lyophilized. Yield 14.9 mg. Deamination of the NH2-terminal Octapeptide-Blood group Mspecific octapeptide (30 mg) and 70 mg of sodium nitrite were dissolved in 3 ml of water, then 1.5 mlof 1N hydrochloric acid was added under cooling (ice bath) and stirring. The mixture was incubated for 30 min at 0 "C and the reaction was terminated by the addition of 60 mgof urea. After 5 min, 100 pl of triethylamine was added and the solution was applied to a column of BioGel P-6 (1.5 x 145 cm), and equilibrated with 0.1 M triethylamine acetate buffer, pH 5.0. The eluate was monitored at 230 nm and the first absorbing peak was pooled and lyophilized (at least three times from aqueous solution). Yield 30 mg. Carbohydrate Analysis-Carbohydrates were determined quantitatively as trifluoroacetyl esters of methyl glycosides by gas-liquid chromatography (Hewlett Packard5708 gas chromatograph equipped with a 5709 electron capture detector and with integrator) as described by Wrann and Todd (15) except for the column. The following conditions were used to identify the sugar derivatives: a glass column (12 f t X 2 mm ID) was packed with HWP 80-100 mesh (1%OV-210), peptides, respectively. argon-methane (95:5) wasused as carrier gas at a flow rate of 60 ml/ min, the temperature was programmed at 120 "C for 4 min, followed MATERIALSANDMETHODS by atemperature increase a t a rate of 2 "C/min to 210 "C and Preparation of Glycophorin A and itsNHz-terminal Tryptic Pep- maintainance of the temperature at 210 "C for 4 min. The injection tides-Glycophorin A was prepared from red blood cells of individuals port was heated at 200 "C and thedetector at 300 "C. Inositol served homozygous for blood group M and N as described by Tomita et al. as an internal standard for the analysis. Response factors were cal(1). The NHn-terminal tryptic peptides T1 and T2 were prepared culated from analyses on pure galactosyl-N-acetylgalactosaminyl[Neither from purified glycophorin A ( 1)or by trypsin digestion of whole acetylneuraminyl]galactosylglucosylceramide ganglioside and its olicells and further purification (13). gosaccharide moiety, both kindly provided by Drs. Laurel 0. Sillerud Preparation of the NH2-terminaE Octapeptides of Glycophorin and Robert K. Yu (Departments of Molecular Biophysics and BioA-NH2-terminal tryptic peptides T1 and T2 were cleaved by cyan- chemistry and Neurology, respectively, Yale University). The carboogen bromide using conditions described by Blumenfeld and Ada- hydratecontent of the glycopeptides is expressed in mol/mol of many (6). Routinely, approximately 200 mg of blood group M- or N- protein after determination of the protein content of an aliquot by specific peptide T1 or T2 was dissolved in 20 ml of 70% formic acid amino acid analysis. Sialic acid was also determined by the thiobarand, after bubbling nitrogen through the solution for 1 min, 6 g of bituric acid method of Warren (16). Amino Acid Analysis-Dry samples of -100 pg of peptide in glass cyanogen bromide (Eastman) was added. The vessel was closed and the mixture was stirred for 35 min a t room temperature in the dark. tubes (7 X 60 mm) were hydrolyzed at 110 "Cfor 24 h under nitrogen The reaction was stopped by at least 30-fold dilution with water and in a sealed glass chamber containing 15 ml of 6 N hydrochloric acid the solution was lyophilized. The peptides were dissolved in 0.1 M with 1%mercaptoethanol and 2% phenol. After hydrolysis, the sampyridine acetate, pH 5.0, or 0.1 M triethylamine acetate, pH 5.0, and ples were dried in a heated desiccator at 65 "C for 1 h under high subjected to gel filtration on a column of BioGel P-6 (1.5 X 145 cm) vacuum and were dissolved in 100 p1of the loading buffer (0.2 M equilibrated with the same buffer. The eluate was assayed for sialic sodium citrate, pH 2.2). The amino acid analyses were performed on acid or monitored for UV-absorbance a t 230 nm, peak fractions were a Durrum D-500 analyzer. Norleucine was used as an internal standpooled, and peptides were recovered by lyophilization (three times ard. The amino acid content of the peptides is expressed in mol/mol from aqueous solution). The second peak contained the octapeptide of peptide. Serological Assays-Blood group M and N activity of peptides (cf. Fig. 1).Yield: -35 mg. Preparation of the NHz-terminal Pentapeptide of Glycophorin was determined by inhibition of hemagglutination as described by A-100mg of blood group MN-specific tryptic peptide mixture T2 Kabat (17). Rabbitantisera specific for blood group M- and Nwas incubated with 5 mgof Staphylococcus aureus V8 protease antigens were purchased from Ortho Diagnostics Inc., Raritan, NJ (Miles) for 100 h at 37 "C in 20 ml of50mM Tris-HC1, pH 7.8 (0.01% (Lot M 198, N 184). Red blood cells of MM and NN phenotype were sodium azide). The mixture was lyophilized, then redissolved in 0.1 obtained from typed laboratory personnel shortly before the assay and were washed three times with phosphate buffered saline (0.85% M pyridine acetate, pH 5.0, and fractionated bygel filtration on a BioGel P-6 column (1.5 x 142 cm) equilibrated with the same buffer. NaC1,0.025 M phosphate buffer, pH 7.2). Glycopeptide solution (25 pl) of known concentration (as deterPooled fractions were lyophilized and furtherpurified by ion exchange chromatography on a DEAE-Sephadex A-25 column (I X 10 em) mined by amino acid analysis) was diluted serially (1:2") with phoswhich had been equilibrated with 2 mM pyridine acetate, pH 5.4. The phate buffered saline on a microtiter plate, and, after addition of 25 p1of appropriately diluted antiserum (four hemagglutinating units), peptide sample was applied in 1 ml of starting buffer and,after washing with an additional 8 ml of starting buffer, the peptides were the mixture was incubated for 30 min at 37 "C; then 25 pl of a 2% (v/ eluted with a linear gradient of pyridine acetate, pH 5.4, from 2 mM v) MM- or NN-erythrocyte suspension was added and theincubation to 1M over a total volume of 100 ml. The pentapeptide was recovered was continued for 1 h at room temperature. Positive agglutination by lyophilization (three times from aqueous solution). Yield: 1.5 mg. was scored + to ++++. The minimal inhibiting concentration was The low yield (20%theoretical) may be explained by the unfavorable calculated from the highest dilution of the glycopeptide solution Glu-Val bond for the VB-protease (14) and steric hindrance by the giving complete inhbition of four hemagglutinating units. I3C Nuclear Magnetic Resonance Measurements-All 13C NMR neighboring bulky oligosaccharide chains. Desialylation of NH2-terminal Octapeptides-Desialylation of spectra were recorded on a Bruker CXP-200 widebore spectrometer blood group M and N specific octapeptides was performed either with at 50.3 MHz. Measurements were made at 20 f 1 "C on 2-ml samples contained in 10-mm NMR tubes fitted with Vortex plugs. Broadband neuraminidase or dilute acid. Octapeptide (20 mg) was incubated with 500 units of neuraminidase proton decoupling was employed throughout. All samples contained (Vibrio cholerae, Calbiochem-Behring) in 2 ml of50 mM sodium approximately 10% D 2 0 which was used as a field-frequency lock. acetate buffer, pH 5.5, with 2 mM calcium chloride and 0.2% toluene Chemical shifts are reported relative to the resonance position of (v/v) for 16 h a t 37 "C. The mixture was lyophilized and the desialy- tetramethylsilane calculated using a chemical shift for internal dioxlated peptide was isolated by gel filtration on a Bio-Gel P-6 column ane of67.86 ppm. The following parameters were typically used: (1.5 x 145 cm) equilibrated with 50 mM acetate acid. Fractions of the spectral width, 10,OOO Hz; data points, 8,192 or 16,384; pulse repetition
Structure of the M N Blood Group Antigens rate, 0.6 to 2.2 s; pulse angle, -40”; digital broadening, 3-5 Hz. When possible, spectra using the longer pulse repetition rate were used for the integration studies. Typical spectra required for 16- to 24-h data accumulation times for peptide concentration ranging from 2 to 5 mM. Quantitative determination of the glycopeptide composition was accomplished by comparison of peak height, computer integration, and cutting and weighing of resolved resonances from expanded plots. In cases where resonances overlapped significantly, the product of the peak height and the width at half-height for the nonoverlapped side of the resonance was used for the comparison. RESULTS
Isolation and Chemical Analysis of NH2-terminal Octapeptides
5783
Isolation of the NH2-terminal Glycosylated Pentapeptide To determine the possible contribution of the COOH-terminal sequence Val-Ala-Hse to the serological activity of the glycosylated octapeptide, the N-specificglycosylated NH2terminal pentapeptide was prepared from a mixture of MNspecific T2 peptide. After cleavage with S. aureus V8 protease, initial chromatography of the digest on a BioGel P-6 column (Fig. 2 A ) , and further purification by ion exchange chromaa pentapeptide could tography on DEAE-Sephadex (Fig. ZB), be isolated with a yield of 1.5 mg/100 mg of starting material, whichwas betterthan 80% pure(Table 11). This peptide apparently was obtained by cleavage at the glutamic acid residue in position 5 of only the N-specific tryptic peptide. This cleavage is consistent with the specificity of the enzyme.
The NHz-terminal octapeptides were prepared as described Modification of NH2-terminal Glyco-octapeptides under “Materials and Methods.” The elution profile of BioGel P-6 gel chromatography was determined by sialic acid assay Desialylation-The blood group M- and N-specific octaand showed three peaks (Fig. l ) , the f i s t of which contained the larger cyanogen bromide fragment of tryptic peptides T1 TABLEI or T2, the second peak contained the NHz-terminal octapepChemical composition of cyanogen bromide derived NH2tide, as determined by amino acid composition, and the third terminal octapeptides of glycophorin A in mol/mol peptide one contained free sialic acid, whichwascleaved off the Amino acid analysis was done ona Durmm D500 amino acid peptide-bound oligosaccharides during the incubation with analyzer. Values of Thr and Ser were corrected for losses during cyanogen bromide in formic acid. hydrolysis. Values for all other amino acids not expected for a given The amino acid composition of the peptides is in accordance peptide were less than 0.05. Carbohydrates were determined by gas with previous studies which suggested differences in protein chromatography. Blood group specificity structure for the two forms with blood group M- and Nactivity. The peptides derived from blood group MN-donor M MN N glycophorin A are mixtures of both (4-6; Table I). The pepThr 2.0 (2)” 2.0 (2) 2.0 (2) tides contain 3 mol of galactose and N-acetylgalactosamine Ser 2.0 (2) 1.5 (1.5) 1.1 (1) each and 6 mol of N-acetylneuraminic acid, suggesting that a Glu (0.5) 0.5 0.9 (1) tetrasaccharide is located at each of the glycosylation sites in GlY 1.0 (1)(0.5) 0.5 Ala 1.0 (1) 0.9 (1) 1.0 (1) positions 2, 3, and 4 of the amino acid sequence consisting of Val (1) 1.0 (1) 1.0 (1) one galactose, one N-acetylgalactosamine, and two N-acetyl- 1.0 Hse 1.0 (1) 0.8 (1) 1.0 (1) neuraminic acid residues. Leu 0.4 (0.5) 1.0 (1) Cyanogen bromide cleavage of the tryptic peptides and Gal 3.1 (3) 2.7 (3) 3.4 (3) recovery of NH2-octapeptides was quantitative. Loss of sialic GalNAc 2.8 (3) 3.3 (3) 2.9 (3) acid under the conditions used was approximately 5% of total NeuNAc 6.1 (6) 5.1 (6) 6.3 (6) NeuNAc* 5.7 (6) 6.3 (6) 5.9 (6) NeuNAc. Whenthe cleavage wascarried out under conditions oflow cyanogen bromide concentration and with 24-h incu“ Expected values in parentheses, based on Structure A, Fig. 3. bation (18),the loss of sialic acid was approximately 20%. * Determined colorimetrically. 1.5
I
FIG. 1 (left). Gel filtration of the cyanogen bromide digest of tryptic peptide T1.The cyanogen bromide-derived peptides of 200 mgof peptide T1 or T2 (not shown) were subjected to gel filtration on a column of BioGel P-6 (1.5 X 145 cm) equilibrated with 0.1 M pyridine acetate, pH 5.0, at a flow rate of 2 ml/fraction/2 min. Eluted fractions were assayed for sialic acid-containing peptides by a colorimetric method (16). NeuNAc corresponds to free sialic acid. FIG. 2 (right). Purification of the NHz-terminal pentapeptide of glycophorin AN.The V8 protease digest of 100 mg ofblood group MN-specific tryptic peptide T2 was subjected to gel filtration on a
column of BioGel P-6 (1.5 X 145 cm) equilibrated with 0.1 M pyridine acetate, pH 5.0 ( A ) .Individual fractions were assayed for sialic acidcontaining peptides by a colorimetric method (16). Peptide material from pool 2 was further purified by ion exchange chromatography on a column of DEAE-Sephadex A-25 (1 X 10 cm) equilibrated with 2 mM pyridine acetate, pH 5.4 ( B ) .A linear gradient was started at the arrow from 2 mM to 1 M pyridine acetate,pH 5.4(100ml, total volume). Flow rate was approximately 10 ml/h, fraction volume, 1.3 ml.
Structure of the M N Blood Group Antigens
5784
peptides of glycophorin A were desialylated either by incubation with neuraminidase or by heating with dilute acid as described under“Materials and Methods.” Neuraminidase had thedisadvantage of contaminating the pure peptide with the enzyme, however, a single gel filtration step (BioGel P-6) was sufficient to recover the pure desialylated peptide as shown in Table 11. Release of sialic acid was better than 90% as judged colorimetrically. Treatment of the octapeptide with dilute sulfuric acid yielded 100% desialylated peptide after a desalting step. The desialylated peptides showed an elution behavior on gel filtration which was altered in comparison to the original peptides and was consistent with a 50% reduction in molecular weight. Deamination-The blood group M-specific octapeptide was treated with nitrous acid to remove the NH2-terminal amino group. Dilute hydrochloric acid was chosen to provide the acidic medium, because diluteacetic acid (2 M) was only partially effective in the same reaction mixture. Under the conditions given under “Materials and Methods,” the NH2terminal aminogroup was destroyed quantitativelyand selectively as judged by amino acid composition (Table 11).In spite of the low pH, there was no loss of sialic acid as determined colorimetrically. Determination of Serological Activity of the NH2-terminal Peptides of Glycophorin A The blood group M- and N-activity of the NH2-terminal peptides of glycophorin A and their derivatives was determined by the hemagglutination inhibition test as described under “Materials and Methods.” For each test, 32 nmol of peptide was used (as determined by amino acid analysis). The concentrations of peptides completely inhibiting four hemagglutinating units are given in Table 111. These datashow unequivocally that theblood group M and N determinants reside within the NHs-terminal octapeptides. Serological activity on a molar basis was found almost quantitatively onthe octapeptides after cleavage of tryptic peptides T1 or T2 with cyanogen bromide and isolation of the products. It is evident that no blood group M and N activity is associated with the large CNBr-fragment of tryptic peptide T1. Furthermore, the dataon the N-specific NH2-terminal pentapeptide prove that it contains the whole blood group N activity and that the COOH-terminal sequence Val-Ala-Hse does not contribute to the activity of the N-specific octapeptide. The results obtainedfrom the modified octapeptides clearly demonstrate the requirement for an intact amino group and sialic acid on the oligosaccharide chains for the expression of blood group M and N activity, as judged by commercial rabbit antisera. TABLEI1 Amino acid composition of NH2-terminalpeptides andderivatives in mol/mol of peptide (see Table I) “octapeptide Untreated
ASP Thr Ser Glu Pro GlY Ala Val Hse Leu
2.0 (2) 2.0 (2) 1.0 (1) 1.0 (I) 1.0 (1) 1.0 (1)
N-octapeptide
2::2.0 (2) 1.0 (1)
1.0 (1) 1.0 (1) 1.1 (1) 0.8 (1)
2.0 (2) 1.1 (1)
2.0 (2)
0.9 (1)
1.0 (1) 1.0 (1)
1.0 (1) 1.0 (1) 1.0 (1) 1.0 (1)
1.0 1.0 (1) 1.0 (1) 1.0 (1)
N-pentapeptide untreated
0.5 1.9 (2) 1.1 (1)
1.1 (1) 0.3
0.9 (1)
I3C-NMRStudies on the NH2-terminal Glyco-octapeptides The analysis of the 13C NMR spectra of the various octapeptides (I-IV, Fig. 3) involves resonance assignments in the case of peptide residues, and both elucidation of the structure and resonance assignments in the case of oligosaccharides residues. Helpful in the following analysis of both peptide and oligosaccharide moieties is noting the presence of four discrete spectral regions in the spectra of I and I1 (Fig. 4), characteristic TABLE111
Serological activity of NH2-terminal peptides of plvcoahorin A ~
Concentration of inhibitors completely inhibiting four hemagglutinating units
Blood group specificity
Inhibitor
Anti“
Tryptic peptide TI Tryptic peptide T2 Octapeptide Larger CB-fragment of TI Pentapeptide Deaminated octapeptide Deaminated, desialylated octapeptide Desialylated octapeptide
Anti-N
nmol/ml 10 >640 >640 IO 10 2640 >640 10
M N M N M N MN N M M
N
. Gly .
-
10
r640
>640 >a0
>640
20
>640
20
>640 >640
>640 >640
>640
>640
Val
. Ala
. Hse
. C02H
. C02H
. Glu
. Val .
Ala , Hse
. Gly
.
Val
.
Ala
.
Hse
.
C02H
.
. Val
.
Ala
.
Hse
.
G02H
Glu
NeuNAc ]a2-6 A, R1 through R 6 ’. NeuNAca2-3GalBJ-3GalNAc 1-Ser or Thr and R ‘ through Ri: GalB1-3GalNAc 1-Ser or Thr
NeuNAc a2-6 R3=R4=R6: NeuNAca2-3GalB1-3GalNAc1-Ser
I
B, R1 = -
or Thr
and
R2
=
R5:
NeuNAcaZ-3GalB1-3GalNAc 1-Thr
and R ‘ through R;:
c, R~ = R
2
=
and
R
4
=
R
5
=
R
3
GalE1-3GalNAc
l-Ser or Thr
: [NeuNA~aGalB]~GalNAcal-Ser or Thr
Gal 15 R ~ :NeuNAcnGalBGalNAcal-Ser o r Thr
and
R‘ through R‘: 6
[Gal61 GalNAcal-Ser o r Thr
FIG. 3. Glyco-octapeptide structures. Top, the peptide structures of M-sialoglycopeptide (I), N-sialoglycopeptide (II), asialo-M (111), and asialo-N (IV).Bottom, proposed structures ( A to C ) for the serine or threonine-linkedoligosaccharides (RIto &) to I and I1 and asialo-oligosaccharides(E1 to R 6 ) of I11 and IV. Proposed structures are A, Thomas and Winder (45);B, Blumenfeld and Adamany (6); and C, Springer and Desai (IO).
Structure of the M N Blood Group Antigens
5785
of the following classes of carbons (19): 15-24 ppm, methyl; 28-78 ppm, methylene and methine; 99-107 ppm, anomeric; and 170-180 ppm, carboxyl and carbonyl.
structural differences that could account for more than one type of NeuNAc residue include differences in linkage site (primary or secondary hydroxyl) and differences in the anomeric form of the glycosidic linkage. Oligosaccharide Structure and Assignments The anomeric form of the NeuNAc residues is assigned by Each of the proposed oligosaccharide structures for I and I1 the presence of resonances near 74 ppm and the absence of involve a repeating structure of one or two distal NeuNAc resonances around 71 ppm, since the former is the characterresidues linked to a proximal or "core" saccharide, composed istic chemical shift of the NeuNAc C-6 carbon in an a-glycoof one or more Gal and GalNAc residues (Fig. 3). In order to sidic linkage, and the latter thesame carbon in a /3-glycosidic establish the oligosaccharide structures of I and 11, one must linkage (20).Thus, it can be concluded that all of the NeuNAc determine: a) the number of NeuNAc residues per repeating residues of I and 11 are a-glycosidically linked to the core structure, b) the NeuNAc linkage sites, and c ) the structure of saccharide and that the different types of NeuNAc residues the core saccharide. These three structural questions can be must result from different linkage sites. It should be noted answered unambiguously through 13C NMR analysis. The that the chemical shift of the anomeric carbons of glycosidifist structural problem considered is that concerning the cally-linked NeuNAc residues are not sufficiently different to be useful in the assignment of the anomeric form in contrast NeuNAc residues. to the distinctly different anomeric shifts used below for Assignments forNeuNAc Residues of Z and IZ assignments of the anomeric form of the Gal and GalNAc The NeuNAc resonances in the spectra of I and I1 (Fig. 4) residues. Determination of the linkage sites of NeuNAc residues were assigned through selective desialization of I and 11 to yield the asialoglycopeptides I11 and IV, respectively. Due to present inI and 11 is accomplished by chemical shift comparthe absence from the spectra of 111 and IV of all the NeuNAc ison ofthe NeuNAc resonances of the two sialoglycopeptides carbon resonances, unequivocal assignment of these reso- with those of two model NeuNAc residues (Table IV). These nances was possible in the spectra of the sialoglycopeptides. model compounds are the a2-6 and a2-3 isomers of N-acetylAs an example, the spectral simplification which results from neuraminyl lactose (21), having the structures NeuNAccu2desialylation of the "active sialoglycopeptide I to form 111 is 6Galpl-4Glc and NeuNAca2-3Gal/31-4Clc,respectively. The shown in Fig. 5, A and B. Through this procedure, 16 reso- former is an analogue of a NeuNAc residue linked to a primary nances were attributable toNeuNAc carbons in the spectra of hydroxyl and the latter is a NeuNAc residue linked to a I and 11. The finding that more than 11 NeuNAc resonances secondary hydroxyl of a Gal or GalNAc residue. Of the 16 are observed for each sialoglycopeptide indicates that more NeuNAc resonances of I and 11, 10 were observed as five sets than one typeof NeuNAc residue is present in each. Possible of paired resonances and six as single resonances. Within 0.2 ppm, and inmany cases much less, one resonance of each pair had a chemical shift identical with a resonance of the a2-6 isomer and the other a chemical shift identical with a resonance of the a2-3 isomer of N-acetylneuraminyl lactose. In general, the upfield member of each pair aligned with a resonance of the a2-6 isomer and the downfield member with the a2-3 isomer. The remaining six resonances from both I and I1 corresponded within experimental error (-t0.2 ppm) to one of the six remaining resonances of either of the two model 1 0
,m
,
,a
.
110
,
,IO
IB
VO
,OD
"'
P
90 10
10
Pr-
-
a
Y)
~
a
10
?Om
FIG. 4. The 13CNMR spectra of the NHZ-terminal,human red cell glycophorin A sialoglyco-octapeptides at 60.3 MHz in 9:l HaO/D20.A, M-immunoreactive sialoglycopeptide(I) at pH 6.8; B, N-immunoreactive sialoglycopeptide (11) at pH 7.5. The resonances around 148 and 138 ppm are due to trace amounts of pyridine. Both spectra wereobtained from -30 mg ofglycopeptide in 2 ml ofsolvent.
$70
160
150
140 Ppn
1 5 0
I20
I10
IO0
'
~
TABLEIV Carbon-13assignments for NeuNAc residues of human glycophorin A NHz-terminalglyco-octapeptides (Z and ZZ) at 50.3 MHz and of model NeuNAc residues for comparison Carbon
1 100.99 2 3 69.63' 4 5 6 69.63' 7 8 9 176.29 10 11
B
IBD
-
80
70
60
50 FQm
40
30
I ("active)," pH 6.8
I1 (N-active),"pH 7.5
N-Acetylneuraminyllactose: pH 7.0
012-3
a24
a2-3
a2-6
a2-3
a2-6
linked
linked
linked
linked
isomer
isomer
174.64
174.64 101.62 41.58 69.63' 53.25 73.88 69.63' 73.25 63.98 176.29 23.43
174.64 101.01 40.97 69.70' 53.06 74.07 69.70' 73.15 64.02 176.29 23.45
174.64 101.64 41.55 69.70' 53.25 73.88 69.70' 73.30 64.02 176.29 23.45
175.0 101.0 40.8 69.4 53.0 74.0 69.4 72.9
174.8 101.6 41.4 69.7 53.2 73.8 69.7 73.1 64.0 176.3 23.4
41.00 53.06 74.05 73.10 63.98 23.43
64.0 176.2 23.4
20
FIG. 5. Effect on the '% NMR spectrum (60.3 MHz) of the Mimmunoreactive glycopeptide resulting from deamination and desialylation. A, M-immunoreactive sialoglycopeptide(I) at pH 6.0.
The resonances around 148 and 138 ppm are due to trace amounts of pyridine. B, deaminated, asialo-M glycopeptide (III) at pH 6.5. The resonance around48 ppm is due to triethylamine. Dioxane is present in both spectra at 67.86 ppm as the internal reference. Both spectra were obtained from-30 mg of glycopeptide in2 ml of solvent.
Estimated accuracy, i0.07 ppm. Chemicalshifts relative to internal dioxane set equal to 67.86 ppm. Spectra were obtained at 20 f 1 OC from 10%D20 solution. Data taken from Ref. 21. Chemical shifts are correctedfor different referencing bysetting NeuNAc-11 equal to 23.4 ppm (+LO ppm). of N-acetylneuraminyl NeuNAc assignmentsare based on comparison lactose spectra with the spectrum of lactose (Ref. 47). Signal integratingfor 15 carbons in each spectrum; assignedto 12 NeuNAc and 3 GalNAc resonances. a
'
5786
Structure of the M N Blood Group Antigens
compounds, except for the pH-sensitive NeuNAc carboxyls TABLEV (k0.4 ppm). Thus, all NeuNAc resonances of I and 11 were Carbon-13 assignments for core saccharide residues of human glycophorin A NH2-terminal glyco-octapeptides(I to IV) at 50.3 assigned (Table IV) to one of two types of NeuNAc residues, MHz one type linked through a primary hydroxyl and the other Estimated accuracy, k0.07 ppm. Chemical shifts relative to internal type linked through a secondary hydroxyl of the Galor dioxane set equal to 67.86 ppm. Spectra were obtained at 20 f 1 "C GalNAc residues of the core saccharide. The observations (Fig. 4) that theintensity of each member from 10% D20 solutions. I ("active, Lu, (asialo- II (N-active, IV (asialo-N, of the five paired NeuNAc resonances is equal (especially pH 6.8) M, pH 6.5) pH 7.5) pH 5.5) number clear around 73-74 ppm as noted by X in Fig. 4A) and that the intensity of the six single resonances approximately equals Gal-1 105.82 105.86 105.91 105.96 the combined intensity of each set of paired resonances indi2 70.31 71.93 70.38 71.89 3 76.96 73.88 76.98 73.88 cate that each type of NeuNAc residue is present in an 4 68.69 69.94 68.83 69.95 equivalent amount. In comparing the areas of the two 5 76.06 76.25 76.06 76.26 NeuNAc-3 carbon resonances near 41 ppm, it should be noted 6 62.40 62.37' 62.42 62.38 that theLeu Cp of I1 at 41.55 ppm must be taken into account 100.67' GalNAc-1 101.01'100.77'100.99' (see peptide assignments below) and that one of the two 100.42" 100.18" 100.18" 100.04" NeuNAc-3 resonances is slightly broader than theother. Thus, 99.46d 99.50d 99.70" 99.65d 49.66 2 49.71 49.70 49.66 it is concluded that there are two types of NeuNAc residues 78.56 3 78.39 78.68 78.59 present in equivalent amounts in both I and 11, one type 69.63' 4 70.19 69.70' 70.24 linked to a primary hydroxyl and the other type linked to70.87 a 5 70.92 72.27 72.33 secondary hydroxyl of the core saccharide. It should be em65.22 6 62.52' 65.09 62.38 phasized that no differences were observed in the 64.70NeuNAc 64.78 resonances of the M and N form of the sialoglycopeptides (I 175.22 7 175.29 175.22 175.29 23.72 8 23.64 23.88 23.79 and II). Further support for the NeuNAc resonance assignments a Deaminated NHz-terminal. * Assignments for these resonances may be interchanged. was obtained from an acid titration experiment of I and 11 in Anomeric resonance assigned to serine-linked GalNAc residue. which it was shown that those resonances assigned to NeuNAc Anomeric resonance assigned to threonine-linked GalNAc resiC-1, (3-2, C-3, and C-8 gave rise to a pH-dependent shift of similar magnitude and direction to those observed in the 13C due. e Signal integrating for 15 carbons in each spectrum, assigned to 12 titration studies' of the a anomer of free N-acetyl neuraminic NeuNAc and 3 GalNAc resonances. acid. The sensitivity of C-8 to protonation has been previously explained to result from the existence of a hydrogen bond assigned to the peptide-linked saccharide residue due to the between the C-8 hydroxyl and the C-1 carboxyl groups (20, expected multiplicity of its anomeric signal, arising from link22). age to both serine and threonine. The signal at 105.9 ppm is therefore assigned to the anomeric carbon of the intersacCore Saccharide Structure of ZII and ZV and Assignments charide linkage. In order to assign the residue order, and The core saccharide structures of 111 and IV were estab- anomeric form of the intersaccharide linkage, the 10 possible lished first since they provide the data base upon which to anomeric shifts must be considered for Gal-1 or GalNAc-1 in calculate the resonance shifts of the more complicated core either al-3, Pl-3,a1-4, or Pl-4 and GalNAc-1 in either al-2 or structures of I and 11. This involved determining a ) the ratio Pl-2 interresidue linkages. Using a consistent referencing sysof Gal to GalNAc residues, b) the linkage order of the saccha- tem of either dioxane at 67.86 ppm or GalNAc-8 at 23.4 ppm, ride residues, c) the anomeric form of each residue, and d ) the these 10 anomeric shifts are obtained from the spectra of galactose oligosaccharides (24),galactan oligosaccharides(25), linkage type between residues. Inspection of thespectra of I11 and IV allows several ganglioside^,^ or Forssman hapten (26) and are as follows (+ carbons of the core saccharide to be readily assigned due to 0.2 ppm): Galal-3 (96.9), GalNAcal-3 (95.8), GalP1-3 (106.2), their intensity and unique presence in characteristic spectral GalNAcPl-3 (104.9), Galal-4 (101.8), GalNAcal-4 (101.4), regions (Fig, 5B). These include Gal C-6 and GalNAc C-6, C- G@1-4 (105.7), GalNAcPl-4 (104.0),GalNAcal-2 (94.8),and 7, and C-8 (Table V). Comparison of the two primary hydroxyl GalNAcPl-2 (103.9). Only twoof the possible anomeric shifts resonances (Gal-6and GalNAc-6) shows each to be of approx- are within the combined experimental error (+ 0.4) of the imately equal intensity and that, in general, all of the more experimental value (105.9 ppm), namely, those arising from intense Gal and GalNAc resonances in the methylene and the GalPl-3 and GalP1-4. Thus, the residue order and anomethine regions are of equal intensity. (see "Integration Stud- meric form of the intersaccharide linkage must beGalPies.") This observation allows one to conclude that the Gal GalNAc and the glycosylated carbon must be either GalNAcand GalNAc residues are present in equivalent amounts and 3 or GalNAc-4. In order to determine specifically which GalNAc carbon is that the linkage between them is not 1-6, since glycosylation of one of the C-6 hydroxyls would have resulted in a downfield involved in the intersaccharide linkage, the chemical shift for both C-3 and C-4 is calculated after glycoside formation to shift relative to that of the primary hydroxyl region (23). Of the remaining structural features of the core saccharide the P-galactosyl residue. For GalNAc-3, the aldose shift (27), to be elucidated, including residue order, linkage type, and 68.9 ppm, should be deshielded 9.4 ppm (23) to 78.3 ppm. For anomeric form of residues, alI can be deduced from the anal- GalNAc-4, the best available model compound (24) predicts ysis of the anomeric signals. Inspection of the anomeric region a chemical shift of 80.1 ppm. In fact, the most downfield signal of I11 and IV reveals an intense signal at 105.9 pprn and a in the methine region of I11 and IV is at 78.4 ppm, almost group of three signals between 101.0 and 99.5 ppm with a exactly the calculated shift for glycoside formation toa combined intensity comparable to that of the resonance at GalNAc-3. Thus, the structure for the core saccharide of 111 105.9 ppm. The three signals between 101.0 and 99.5 ppm are and IV must be Galpl-3GalNAc. Sillerud, L. O., Prestegard, J. H., Yu, R. K., Schafer, D. E., and Sillerud, L. O., Prestegard, J . H., Yu, R. K., Schafer, D. E.,
azt$Gn
Konigsberg, W. H., unpublished data.
Koerner, T. A. W., Jr., and Konigsberg, W. H., unpublished data.
Structure of the M N Blood Group Antigens
5787
The anomeric form of the peptide-linked GalNAc residue ride units of the antifreeze glycoprotein support our structure can bededuced by comparison of its chemical shift with and assignments for the threonine-linked core saccharides of calculatedshifts, as doneabove for the Gal residue. For 111 and IV. GalNAc-1 in a Dl-0-Ser linkage, the expected chemical shift Core Saccharide Structure of I and 11 and Assignments is 103.9 ppm calculated from the chemical shift of the parent Having established the structure of the core saccharide of &aldose (27), 96.7 ppm, corrected with +7.0 ppm for methyl glycosidation (28) and +0.2 ppm for methyl + serine substi- the asialoglycopeptides, the Gal and GalNAc carbons of the tution (29). Based on a -2.1 ppm effect for aglycon substitu- intact sialoglycopeptides I and I1 can be assigned if the tion (30), GalNAc-1 in the pl-0-Thr linkage is expected at NeuNAc linkage sites are known. All possible linkage sites 101.8 ppm. For GalNAc-1 in an a1-0-Serlinkage, the expected must be considered due to the existing controversy (Fig. 3). chemical shift is 101.2 ppm, calculated from the chemical shift These sites can be established by comparison of calculated of the parent a-aldose (27), 92.4 ppm, corrected with +7.5 with experimental shifts inthe spectra of I and I1 as follows. ppm for methyl glycosidation (28), and +1.3 ppm estimated Six carbons of the Gal and GalNAc residues of 111 and IV are for methyl + serine substitution (31). Based on a -1.7 ppm available a priori as NeuNAc attachment sites. Using the effect for aglycon substitution (30), GalNAc-1 in the a1-0-Thr chemical shifts of each of these six carbons (Table V) and the linkage is expected at 99.5 ppm. Comparison of these calcu- known values (21) for the deshielding induced by sialylation lated ranges for p (103.9-101.8 ppm) and a (101.2-99.5 ppm) of a primary hydroxyl (+2.5) and of a secondary hydroxyl linkage of GalNAc-1 to serine and threonine with the experi- (+3.0) of a Gal or GalNAc residue, the calculated chemical mental chemical shifts (101.0-99.5 ppm) clearly shows the shifts for the six possible sialylated hydroxyl carbons are as GalNAc residue of I11 and IV are a-0-linked to serine or follows: Gal-2, 74.9 ppm; Gal-3, 76.9 ppm; Gal-4, 72.9 ppm; threonine. Thus, the structure of the core saccharide is finally Gal-6, 64.9 ppm; GalNAc-4, 73.2 ppm; and GalNAc-6, 65.0 ppm. Inspection of the spectra of I (Fig. 5A) and I1 reveals concluded to be Galpl-3GalNAcal-0-Ser or Thr. Elucidation of the core saccharide structure of 111 and IV two sialylation-induced changes in the glycoside region, allows the chemical shifts of the remaining carbons of Gal and namely a single resonance at 77.0 ppm and a closely spaced GalNAc to be calculated using known glycosidic substituent pair around65 ppm, the paired resonances having a combined effects and the chemical shifts of model methyl glycosides area comparable to that of the single resonance. Since the (Table VI). Since there is close agreement between the cal- resonances near 73.0 ppm have previously been assigned to culated and observed shifts, the signal assignments for these the NeuNAc residues, I and 11 must contain two NeuNAc linkages, one to Gal-3 and the other to either Gal-6 or GalNAcremaining carbons was straightforward (Table V). Recently,Berman et al. (32) havereporteda I3C NMR 6. In order to determine whether Gal-6 or GalNAc-6 bears a study of antifreeze glycoproteins 3 and 6 from the Antarctic NeuNAc residue, both possible C-5 chemical shifts were calcod which contain the known repeating structureGalpl3GalNAcal-0-Thr. Since we have assigned the same structure culated for the two residues, using the shifts for Gal-5 and to the core saccharide of two of thethree NHp-terminal GalNAc-5 in 111 and IV (Table V) and the known Ay-glycooligosaccharides of glycophorin A (111 and IV), it was of sylation shift (21) for a carbon penultimate to sialylation of a interest to compare the chemical shifts reported by Berman primary hydroxyl group (-1.3 ppm). Thecalculated values for et al. (32) for the Gal and GalNAc carbons of this disaccharide Gal-5 and GalNAc-5 are 75.0 and 71.0 ppm, respectively. structure with those we have reported (Table V). This com- Inspection of the spectrum of I and 11 reveals no resonance at parison shows the chemical shift difference of all carbons is or near 75.0 ppm, but a resonance is observed at 70.9 ppm. 0.2 ppm or less, with the exception of GalNAc-3 (-0.5 pprn), Thus, two NeuNAc linkages exist, one to Gal-3 and the other GalNAc-6 (-0.4 pprn), and GalNAc-7 (-1.5 ppm). Theslightly to GalNAc-6. This finding is consistent with our previous greater chemical shift difference of GalNAc-3 and -6 are conclusion that the multiplicity of the NeuNAc resonances probably due todifferences in pH (0.5 to 1.0) and temperature was due to the presence of both a primary and secondary (15 "C). The large difference for GalNAc-7 is dueto the hydroxyl-linked NeuNAc residue in I and 11, both present in extreme temperature sensitivity of the chemical shift of this equivalent amounts. carbonyl carbon (32). Thus, theassignments for the disacchaHaving thus demonstrated that each oligosaccharide of the TABLEVI Glycosylution shifts (A) of I to IV and model compounds for comparison Carbon
Glycosylation shift"
EffectSubstituent
GalP1-3
AY
A7
GalNAc-2 GalNAc-3 GalNAc-4 Gal-2 Gal-3 Gal-4 GalNAc-4 GalNAc-5 -1.4
AI'
GalNAc-6
Alj
NeuNAca2-3
A' Ar A0 A7
NeuNAca2-6
AS
I
Ill
-1.5 +9.7
-1.5 +9.7 +0.4
d
~~
- 1.6 +3.1 -1.3 -0.6 -1.4 (+2,3 +2.7
" Galactosylation shifts are calculated as follows: TI1 or IV minus T h e values for the methyl 2-acetamido-2-deoxy-a-galactopyranoside. latter are taken as:C-2, 51.2; C-3, 68.9; and (2-4, 69.8 ppm, calculated from aGalNAc (Ref. 27) and the methylglycosidation shifts of methyl 2-acetamido-2-deoxy-a-glucopyranoside (Ref. 28). Sialylationshifts are calculated as follows: I minus I11 or I1 minus IV. T h e model galactosylation shifts are calculated asfollows: model
'
I1
Iv ..
-1.5 +9.8
-1.5 +9.7 +0.4
"
rl
-1.5 +3.1 -1.1 -0.5 +2.7 +2.31
glycO-
shift* sylation
Model oligosaccharide"
-0.8 GalP1-3 L RhaP +9.4 GalP1-3 L RhaP -0.1 GalP1-3 L RhaP -1.6 NeuNAca2-3 Galpl-4Glc +3.0 NeuNAca2-3 Gal/ll-4Glc -0.9 NeuNAca2-3 GalPl-4Glc -0.8 NeuNAca2-6 Galpl-4Glc -1.3GalPl-4Glc NeuNAca2-6 +2.5 GalPl-4Glc NeuNAca2-6
minus PL-Rha, the values for the latter taken from Ref. 23. Model sialylation shifts are calculated as follows: model minus lactose, the values for the latter taken fromRef. 47. Values for Galpl-3~-RhaB taken from Ref. 23; values for a2-3 and a2-6 isomers of N-acetylneuraminyl lactose taken from Ref. 21. Galactosylation shift combined with sialylation shift.
5788
Structure of the M N Blood Group Antigens
TABLEVI1 sialoglycopeptides contains Gal-3 and GalNAc-6 linked NeuNAc residues and having previously shown that thecore Carbon-13 assignments for the amino acid residues of the NH2terminal glyco-octapeptides (I to IV) at 50.3 MHz saccharide has the structure GalPl-3GalNAcal-O-Ser(Thr) Estimated accuracy. k0.07 m m . Chemical shifts relative to internal and all NeuNAc residues are a-glycosidically linked, it can only be concluded that the oligosaccharides of I and 11 have dioxane set qual to 67.86 ppm. Spectra were obtained at 20 k 1 "C from 10% 1 1 solutions. the structure NeuNAca2-3Gal~l-3(NeuNAca2-6)G~NAc~lPosition, 0-Ser(Thr). Thisstructure is identical to structure A (Fig. 3); residue and I ("active, IV (asialo-N, however, our structure is more complete since we have spec- carbon despH 6.8) 7.5) pH 6.5) pH pH 5.5) ignation ified the anomeric form of the GalNAc-peptide linkage. With the structures of I and I1 established, the chemical shifts of 1. Ser C" C" 56.01 59.85' sialylation-sensitive Gal and GalNAc carbons can be calcuCO 62.40 62.37',' lated using known sialylation shifts (Table VI). These calcuLeu C" lated shifts all agree well with observedvalues,allowing C" 54.27 54.17 assignments for all core saccharide resonances of I and I1 to C" 41.55 41.26 be made (Table V). CY 1II 25.20 25.05
Peptide Residue Assignments The amino acid sequence of the NHz-terminal region of the "active and N-active forms of glycophorin A molecules is well established. With the exception of the carbonyl carbon resonances, all remaining peptide carbon resonances are expected (19) to occur in the region 15-70 pprn (Fig. 4 and 5) and to be much less intense than the resonances from the repeating oligosaccharides. Due to limited sample quantity, no attempt was made to assign the peptide carbonyl resonances. Assignment of all other peptide resonances is presented in Table VI1 and is based on the following arguments. All carbons of the alanine and valine residues of I-IV, glycine of 1 and 111, and glutamate of I1 and IV are distant from the polypeptide terminals and oligosaccharide attachment sites and are thus readily assigned on the basis of a comparison of their chemical shifts with those of the terminalblocked tetrapeptide and other model peptides containing these amino acids (33-35). The chemical shifts of the carbons of the glycosylated serine and twoglycosylated threonine residues were assigned by comparison with calculated values since all of the necessary model compounds have not been reported in the literature. These calculations were based on the recently reported (29) chemical shifts for the C" and Cp serine carbons of GalNAcpl-0-Ser, 56.3 and 69.2 ppm, respectively. From these chemical shifts and the chemical shifts of the C" and C" carbons of serine (29), the glycosylation shifts were calculated: AC" = -1.8 and ACp = +7.4 ppm. Applying these glycosylation shifts to the known shifts for the C" and Cp carbons of intrapeptide serine (33), 56.6 and 62.3 ppm, respectively, and correcting for the change in anomeric form (Ref. 30; Ace = -0.3, ACp = -1.7) yields 54.5 and 68.0 ppm, respectively, as the calculated chemical shifts for the C" and CO carbons of aGalNAc-linked serine residues of I-IV. Experimentally, one observes resonances at 53.1-53.8 and 68.5-68.8 ppm in the spectra of I-IV which are thus assigned to the C" and Cp carbons, respectively, of the glycosylated serine residues of I-IV. In thecase of the glycosylated threonine carbons, the above glycosylation shifts are applied to the known shifts for the C" and CDcarbons of intrapeptide threonine (33), 60.2 and 68.3 ppm, respectively, and correcting for the change in anomeric form (Ref. 30;AC" = 0.2,ACp = -1.1) yields 58.2 and 74.6, respectively, as the calculated chemical shifts for the C" and Cfi carbons of aGalNAc-linked threonine residues of I - N . Experimentally, two resonances of equal intensity are observed in the spectra of I-IV at 58.0-58.3 and 58.5-58.7 ppm. We have assigned both these resonances to the C" carbons of the two threonine residues of I-IV. In the spectra of I and II, the only resonance which is observed in the region of74.6 ppm, predicted for the Cp carbon of threonine, has been previously assigned to C-6 of the NeuNAc residue and this
CS C
2. Ser C"
C" CU
:
I t
22.67 23.45'
22.62 23.06
53.06 68.69
53.78 68.59
53.06 68.83
53.11 68.53
58.28d 80.02' e 19.79'
58.30" 77.61 19.61'
58.06 80.04', 19.95'
57.96 77.91 19.57'
58.5gd 80.02' ' 19.79'
58.50" 77.61 19.61'
58.67 80.04'. 19.95'
58.59 77.91 19.57'
42.85
42.86 55.07 29.95 34.73
55.00 29.57 37.80
3. Thr C"
c" CO CY 4. Thr C"
C" CY 5. Gly C"
C(' Glu C" C" I CU I CY CS 6. Val C"
c" CU CY CY'
I
60.69 31.41 19.11 19.79'
60.58 31.45 18.93 19.61"
61.11 31.07 19.52 19.95'
60.82 31.17 19.57' 19.57'
51.12 17.67
50.97 17.67
51.22 17.67
51.02 17.67
7. Ala C"
C" CO 8. Hse C"
14.74 (50.37)' i4.67 (50.35)' 15.53 (29.22)' 15.39(29.08)' 19.90 (67.45)' i9.80 (66.40)'
Resonances in region 170-177 ppm.
' Resonances of L-glyceryl residue, resulting from deamination of NHs-terminal serine. e Signal containing resonances of multiple carbons. Assignments may be interchanged. e Tentative assignment. 'Assignments in parentheses are for the lactone form of homoserine. The relative amount of free and lactone forms of homoserine varied with pH.
same region of the spectrum is devoid of resonances in the asialoglycopeptides I11 and IV. On the other hand, an unassigned resonance of appropriate intensity is observed at 77.6 and 77.9 ppm in the spectrum of I11 and IV, respectively, and these have been assigned to thethreonine Cp carbon. Returning to the spectra of I and II, a resonance is observed at 80.2 ppm, which is the only unassigned resonance near 78 ppm, and therefore we have assigned this resonance to thethreonine Cp carbon. The methyl carbons of the two glycosylated threonine residues in I-IV have been assigned to the intense resonance at 20 ppm. With the exception of the 80 ppm resonance, which we have assigned to the Cp carbon of I and 11, the resonance assignment for all three threonine carbons
Structure of the M N Blood Group Antigens
5789
TABLEVI11 Observed and calculated residue ratios for glycopeptides Z to IV Observed“ Residues and carbons compared
1.2
Gal-B/GalNAc-P Gal-5jGalNAc-5 2.9Gal-G/Glya or GluP 3.3
I d
Ill
1.2
Calculatedh
I1
d
IV
Structure A‘ I, 11, Ill, 1v
Structure E‘ 1, 11, 111, IV
Structure C‘ I, 11, 111, 1v
1.1 1.1
1.o
1.0
1.0 1.0 3.0
2.0 2.0 6.0
I. I1
I. I1
3.3
I
I1
2.0 1.7 2.1 2.3‘ NeuNAc-Il/GalNAcB 2.0 2.0 1.7 2 .O‘ 2.2 NeuNAc-5/GalNAc-2 1.0 0.5 2.0 1.7 2.1 1.9’ NeuNAc-S/Gal-B of glycopeptide, as predicted in Fig. 3. per mol Obtained as the ratio of the integrations of well resolved individual Oligosaccharide structures A to C from Fig. 3: Structures A, Ref. carbon resonances fromthe two residues to be compared. Methodsof integration are as described under “Materials and Methods.” Esti- 45; B , Ref. 6; and C, Ref. 10. One resonance of pair was not resolved. mated accuracy, f 0.2.Resonancesused to compute a given ratio ‘Values maybeslightlyhigher due to the presenceoftrace were of the same protonation. * Ratios are calculated from the number of moles of each residue amounts of hydrolyzed NeuNAc residue. 2.0
1.O 1.o
~~~
are in excellent agreement with the recently published (32) only oligosaccharide structure A is consistent with all of the assignments for the glycosylated threonine carbons of anti- 13C NMR data. freeze glycoproteins. Calcium Titration Studies The carbons of the NHB-terminal serine of I and I11 were assigned by their sensitivity to acid titration, the change in In view of the six NeuNAc residues present in theM- and their chemical shift after nitrous acid deamination, and by N-active sialoglycopeptides and theknown propensity of free comparison with the spectrum of L-glyceric acid. The carbons NeuNAc to complexcalciumions (38, 39), we decided to of the COOH-terminal homoserine residue of I-IV were as- investigate by 13C NMR the possible interaction of calcium signed by comparison with the chemical shiftof the carbons ion with the intact“active octapeptide (I).A sample of I at by their sensitivityto acid titration. pH 6.0 was titrated with 0, 6 , and 12 eq of calcium chloride of the free amino acid and It was expected (36) and observed that resonances due to the and the I3C spectrum recorded a t each increment of calcium. lactone form of homoserine were present. The assignment of Only at a calcium concentration of 12 mol eq could we detect, the carbons of the glutamate and NH2-terminalleucine of I1 in the computer-generateddifference spectrum, a small (s0.2 and IV were also supported by demonstration of their acid ppm) but significant chemical shift differencein the resosensitivity. The acid titration study of the tripeptide Leu-Ser- nances which had been previously assigned to the NeuNAc Glu (37) provided additional support for the NHn-terminal carbons 3,4,7,9,10, and 11. At this stoichiometryof calcium, leucine residue assignments of I1 and I V . no calcium-induced resonance shiftsof other carbonscould be detected in thedifference spectra. Integration Studies The identity and magnitude of the resonance shiftsfor the particular NeuNAc carbons observed here bears little resemWith the assignments of all the oligosaccharide and peptide carbon resonances complete, it is possible to make quantita- blance to the reported shifts for free NeuNAc (38, 39) a t a tive comparison between the intensities of specific carbon ratio of calcium to NeuNAc of 2:l which would be representative of our system. These very preliminary results do not resonances and to compare the resulting ratios with those predicted from the threeproposed oligosaccharide structures provide sufficient data toallow us to define the nature of the of the core saccharide interaction; however, we can conclude that there is an interin Fig. 3. From an initial consideration structure,inspection of Fig. 3shows that oligosaccharide action of calciumion withthe sialoglycopeptide I andit structures A and B predict a Gal/GalNAc residue ratioof 1.0 appears to be specifically with the NeuNAc residue(s). and a Gal/peptide residue ratio of 3.0; whereas structure C DISCUSSION predicts a Gal/GalNAc ratio of 2.0 and a Gal/peptide ratio of Previous work on the structure of the M- and N-antigenic 6.0. In fact, integration of well resolved carbon resonances of I-IV (Table VIII) gives Gal/GalNAc ratios of 1.1 2 0.2 and determinant, predominantly from Springer’s laboratory, has Gal/peptide ratios of 3.1 f 0.2. Thus, based on13Cintegration emphasized the carbohydrate natureof the antigenic site(10, C 11).It has been proposed that sialic acid residues contribute studies as well as the chemical shifts assignments, structure can be eliminated from further consideration. in important ways to antigenic activity and thata difference In order to differentiate between structures A and B,it is in carbohydrate content, the number of sialic acid residues, necessarytodeterminetheNeuNAc/GalandNeuNAc/ and oligosaccharide structure of the M- and N-form of the GalNAc ratios, since structure A predicts these ratios to be major sialoglycoprotein (glycophorin A), determinesM- or N2.0 and structure B predicts 1.7. As seen in Table VIII, both blood groups on the surfaceof the human erythrocyte. It has of these ratios are observed in the spectra of I and I1 to be 2.1 been postulated further by this group that the so-called T+- 0.2. The range of values for this ratio introduces some antigen found on erythrocytes after treatment with neuramincells (11) isthe incompletely uncertainty in making a clear distinction between the two idaseandoncertaintumor proposed structures; however, we are confident that the av- glycosylated biosyntheticprecursor of theMN antigens. erage ratio of 2.1 is significantly different from 1.7 and there- There are,however, a number of inconsistencies which do not fore structure B can be excluded. Furthermore, the existence warrant theseconclusions. It has beenfirmly established (and of structure B would give rise to a 2:l multiplicity in the additional support is provided here) that glycophorin A molresonances of the GalNAc residues with large differences in ecules isolated from red cells of individuals homozygous for thechemicalshiftsfromits C-4, C-5, and especially C-6 M or N have a distinctamino acid sequence.TheNH2carbons. None of this is observed. Thus, it is concluded that terminal serinein glycophorin AMis substituted for leucine in
5790
Structure of the M N Blood Group Antigens
ANand the glycine in position five for aglutamic acid residue. Serological activity appears to be strictly correlated with this difference in primary structure, since glycophorin B found in membranes of all human erythrocytesregardless of MN-blood group antigens expresses N-activity and has an amino acid sequence which is identical to glycophorin AN for the NHzterminal 22 residues (4).The possible contribution of the NHzterminal residue, which was confirmed here, had been suggested earlier on the basis of modification with aminoreactive compounds (40) and more recently, by Lisowska and Wasniowska (41) by specific modification of the NH2 terminus in small peptides. antigenic There is no absolute requirement for sialic acid for activity, since some antisera will still distinguish between M and N after removal of sialic acid (9, 42). Nevertheless, desialylation has been shown to destroy the blood group activity measured with many antisera including the sera we have tested. Different antibody specificities and structuralrequirements of the various antisera for binding to the MN determinants was suggested from sialyltransferase experiments (43).After desialylation of intact red cells, enzymatic transfer of sialic acid residues by transferases specific for the linkages to Gal-3 or GalNAc-6 could restore MN activity. Some antisera required the structure NeuNAca2-3Gal, others required NeuNAca2-GGalNAc, and some required both NeuNAca23Gal and NeuNAca2-6GalNAc. From the chemical data presented here, it is also clear that no carbohydrate differences exist within the NHderminal cluster of three oligosaccharideswhich couldcontribute to the difference in antigenic activity between M and N serotypes. It would thus follow that thepreviously mentioned T-antigen is not a precursor for the MN antigens since T-antigen is detectable on glycophorin A regardless of the MN type. This also implies that the T-antigen found on certain tumor cells may bea so-called “tumor antigen,” but certainly its presence is not by itself indicative of the presence of glycophorin A in the membrane of these cells. The expression of the MN antigen as we demonstrate here essentially requires a particular polypeptide backbone. We have extensively analyzed the MN-active sialoglycooctapeptides I and I1 and their derivatives 111 and IV by I3C NMR to substantiate their chemical analysis and to address the question of whether the differences in primary structure of I and I1 are correlated with different conformational arrangements of the oligosaccharides. Specificallly, we asked the question whether thereare interactions of sialicacid carboxyl groups with terminal amino groups. 13CNMR studies have previously been reported only for glycoproteins (32, 44) and glycosylated amino acid (29) devoid of sialic acid. By the various procedures discussed under “Results” we were able to completely assign all of the resonances of the sialoglycopeptides I and II. Both chemical and I3C NMR analyses demonstrate that there aretwo threonine-linked and oneserine-linked tetrasaccharides in the “active (I) and N-active (11) octapeptides from glycophorin A with the identical structure NeuNAca23Galfll-3(NeuNAccu2-6)GalNAcal-O-Ser or Thr. This tetrasaccharide structure was first proposed by Thomas and Winder (45) and hasrecently been c o n f i i e d (46).However, these earlier studies did not characterize the anomeric form of the GalNAc-peptide linkage which we have shown to be a. Previously, the anomeric form of this linkage couldonlybe inferred from differential enzymatic activities (46). We did not find trisaccharides or smaller saccharides within the NHP-terminal region of I and 11. Tri- and disaccharides have been reported in studies on oligosaccharides released through alkaline degradation of native sialoglycoproteins or
larger glycopeptides (45, 46), and one trisaccharide and two tetrasaccharides were proposed to exist within the NHz-terminal region of glycophorin A (6). It is likely that some of these differences are due to hydrolysis (45, 46) but di- and trisaccharide units, nevertheless, may exist further along the polypeptide chain as suggested from compositional data. The I3C NMR analysis c o n f i s the peptide structures of I and 11 as previously reported (4-6). Thus, the only difference in the primary structure of both antigenically active forms is the peptide structure at positions one and five. Several features of the I3C NMR spectra of I-IV deserve discussion since they may provide clues astoa possibly distinct secondary structure of these glycopeptides. First, it is noted in Table VI1 that sialylation of I11 and IV (to form I and II, respectively) causes a deshielding of the C” and Cp carbons of the two threonine residues as well as one of the two valine methyls (CY’).Such a deshielding could not be due to a through-bond effect since there are at least five bonds between the affected carbons and the nearest sites of sialylation. Second, the GalNAc-6 resonance displays a multiplicity in the spectra of I and 11 not seen in the spectra of I11 and IV. Third,the methyl carbons of the two threonine ( C y ) and valine residue (CY’) display a significant pH-sensitivity (0.20.4 ppm). Allof the above noted features may be explained if it is assumed that a glycopeptide secondary structure exists (Fig. 6) that is stabilized by a maximum separation of the three oligosaccharidemoieties. Construction of a space-fiing model with maximally separated oligosaccharide groups leads to the ready formation of three hydrogen bonds involving 11 atom circuits, each between a GalNAc-6 hydroxyland the amide of the next peptide residue in the COOH-terminal direction, for example, the 2Ser GalNAc-6hydroxyl bonds to the3Thr amide. The protons of each of these three hydrogen bonds is
1
HOCH, ACNk
I:
R I = HOCH,, R3.H
I[:
R , = (CH3J2CHCH2,R3=C02CH2CHz
0
IdI: R 2 = ( N e u N A c a 2 - 3 Gal B l - 3 ) c 1’1: R,= NH(CH3) CH CONH(HOCH2CH2)CH COP
FIG.6. Proposed glycopeptide secondary structure. Circled numbers refer to peptide residue sequence positions. The polypeptide backbone is highlighted for reference. In addition to the previously proposed intraresidue NeuNAc hydrogen bonds (20, 22) four hydrogen bonds are shown, three between GalNAc-6 hydroxyl group oxygens and the amide hydrogens of the next peptide residue, and one between the NH2-terminalprotonated amino group and the next peptide residue carbonyl oxygen. The three GalNAc/peptide hydrogen bonds are stabilized by the maximum separation of the negativelycharged NeuNAc residues. The possible glycopeptide tertiary structure resulting from this secondary structure is discussed in text.
Structure of the M N B lood Group Antigens
5791
Sci. U. S. A. 75,2727-2731 7. Furthmayr, H., Metaxas, M., and Metaxas-Buhler, M. (1981) Proc. Natl. Acad. Sci. U. S. A. 78, 631-635 8. Springer, G. F., and Ansell, N. J. (1958) Proc. Natl. Acad. Sci. U. S. A. 44,182-189 9. Judd, W.J., Issitt,P. D., Pavone, B. G., Anderson, J., and Aminoff, D. (1979) Transfusion 19, 12-18 10. Springer, G. F., and Desai, P. R. (1974) Biochem. Biophys. Res. Commun. 61,470-475 11. Springer, G. F., Desai, P. R., Murthy, M. S., Tegtmeyer, H., and Scanlon, E. F. (1979) Prog. Allergy 26.46-96 12. Dahr, W., Uhlenbruck, G., Janssen, E., and Schmalisch, R. (1977) Hum. Genet. 35,335-343 13. Furthmayr, H. (1978) J. Supramol. Struct. 9, 79-95 14. Houmard, J., and Drapeau, G. R. (1972) Proc. Natl. Acad. Sci. U. S. A. 69,3506-3509 15. Wrann, M. M., and Todd, C. W. (1978) J . Chromatogr. 147,309316 16. Warren, L. (1959) J. Biol. Chem. 234, 1971-1975 17. Kabat, E. A. (1956) Blood Group Substances, Academic Press, New York 18. Tomita, M., and Marchesi, V. T. (1975) Proc. Natl. Acad. Sci. U. S. A. 72,2964-2968 19. Rosenthal, S. N., and Fenler, J. H. (1976) Adu. Phys. Org. Chem. 13,379-424 20. Jennings, H. J., and Bhattacharjee, A. K. (1977) Carbohydr. Res. 55, 105-112 21. Jaques, L. W., Glant, S., and Weltner, W. (1980) Carbohydr. Res. 80,207-211 22. Bhattacharjee, A. K., Jennings, H. J., and Kenny, C. P. (1978) Biochemistry 17,645-651 23. Colson, P., and King, R. R. (1976) Carbohydr. Res. 47, 1-13 24. Cox, D. D., Metzner, E. K., Cary, L. W., and Reist, E. J . (1978) Carbohydr. Res. 67,23-31 25. Usov, A. I., Yarotsky, S. V., and Shashkov, A. S . (1980) Biopolymers 19,977-990 26. Koerner, T. A. W., Jr., Li, Y.-T. and Li, S.-C. (1981) Biochem. J., in press 27. Bundle, D. R., Jennings, H. J., and Smith, I. C. P. (1973) Can. J. Chem. 51,3812-3819 28. Perkins, S. J., Johnson, L. N., Phillips, D.C., and Dwek, R. A. (1977) Carbohydr. Res. 59, 19-34 29. Dill, K., and Allerhand, A. (1979) FEBS Lett. 107,26-29 30. Tori, K., Seo, S., Yoshimura, Y., Arita, H., and Tomita, Y. (1977) Tetrahedron Lett. 179-182 31. Grasdalen, H., and Painter, T. (1980) Carbohydr. Res. 81, 59-66 32. Bennan, E., Allerhand, A,, and DeVries, A.L. (1980) J. Biol. Chem. 255,4407-4410 33. Grathwohl, C., and Wuthrich, K. (1974) J . Magn. Reson. 13,217225 34. Keim, P., Vigna, R. A., Marshall, R. C., and Gurd, F. R. N. (1973) J. Biol. Chem. 248,6104-6113 35. Keim, P. Vigna, R. A., Morrow, J . S., Marshall, R. C., and Gurd, F. R. N. (1973) J. Biol. Chem. 248, 7811-7818 36. Liu, T.-T. (1977) in The Proteins (Neurath, H., Hill, R. L., and Boeder, C., eds) Vol. 111, pp. 239-402, Academic Press, New York 37. Gurd, F. R. N., Lawson, P. J., Cochran, D. W., and Wenkert, E. (1971) J. Biol. Chem. 246,3725-3730 38. Jaques, L. W., Brown, E. B., Barrett, J. M., Brey, W. S., and Weltner, W. (1977) J. Biol. Chem. 252,4533-4538 Acknoculedgments-We acknowledge the expert technical assist- 39. Czarniecki, M. F., and Thornton,E. R. (1977) Biochem. Biophys. Res. Commun. 74, 553-558 ance provided by B. Procter and G. Davis for the protein chemical part of this work and thank Dr. Laurel 0. Sillerud for his interest and 40. Lisowska, E., and Duk, M. (1975) Eur. J. Biochem. 54,469-474 41. Lisowska, E., and Wasniowska, K. (1978) Eur. J . Biochem. 88, advice throughout the studies. 247-252 REFERENCES 42. Sturgeon, P., McQuiston, D. T., Taswell, H. F., and Allan, C. J. (1973) Vox Sang. 25,481-497 1. Tomita, M., Furthmayr, H., and Marchesi, V. T. (1978) Biochemistry 17,4756-4770 43. Sadler, J. E., Paulson, J. C., and Hill, R.L. (1979) J. Biol. Chem. 254,2112-2119 2. Marchesi, V. T., Furthmayr, H., and Tomita, M. (1976) Annu. Rev. Biochem. 45,667-698 44. Dill, K., and Allerhand, A. (1979) J . Biol. Chem. 254,4524-4531 3. Furthmayr, H. (1981) in Biology of Complex Carbohydrates, 45. Thomas, D.B., and Winzler, R. J . (1969) J. Biol. Chem. 244, 5943-5946 (Ginsburg, V., ed) Wiley & Sons, New York, in press 4. Furthmayr, H. (1978) Nature 271,519-524 46. Lisowska, E., Duk, M., and Dahr, W. (1980) Carbohydr. Res. 79, 5. Wasniowska, K., Drzeniek, Z., and Lisowska, E. (1977) Biochem. 103-113 Biophys. Res. Commun. 76, 385-390 47. Pfeffer, P. E., Valentine, K. M., and Parrish, F. W. (1979) J . Amer. 6 . Blumenfeld, 0. O., and Adamany, A. M. (1978) Proc. Natl. Acad. Chem. SOC.101, 1265-1274
donated by the amide moiety, leaving the GalNAc-6 hydroxyl available for substitution. The effect of the presence of these three hydrogen bonds is that upon sialylation of the GalNAc6 hydroxyl, the NeuNAc carboxyl group is juxtaposed to the two threonine and valine residues. Since in the model the orientation of the C" and Cpcarbons of the two threonine and the Cy' carbon of the valine are equatorial to theplane of each NeuNAc carboxyl group, the deshielding of these carbons is explained. Moreover, the pH-sensitivity of the threonine and valine methyl carbons is reasonable due to their proximity to NeuNAc carboxyl groups. The slightly different orientation of the NeuNAc carboxyl group when juxtaposed to threonine as compared to thevaline residue can explain the multiplicity of the nearby GalNAc-6 carbon after glycopeptide sialylation. Another interesting spectral feature is the deshielding of the 'Ser C" carbon when the NHz-terminal amino group of I is deaminated (Table VII). This deshielding is probably not a through-bond effect since it occurs three bonds from the site of deamination. However, a hydrogen bond involving an eightatom circuit readily forms in the space-filling model of the unmodified glycopeptide between the 'Ser protonated amino and the 'Ser peptide carbonyl. Thus, deamination could lead to destruction of this hydrogen bond and the deshielding of the nearby 2Ser C" carbon. Taken together, the effects of the above noted possible hydrogen bonds would beto form a tertiary structurein which the polypeptide chain forms an inner core with the NHzterminal in close association with the3Thr a2-6 NeuNAc residue. Spiraling down the polypeptide core approximately 120" apart would be the other oligosaccharide groups. An importantfeature of thistertiarystructure wouldbe the grouping of the NHz-terminal residue ('Ser for I and Leu for JI) and one NeuNAc residue (3Thra2-6)much above the rest of the structure. Thus, the immunodominance of the NH2terminal amino acid residue and the requirement for sialylation could be rationalized in terms of the greater availability and close proximity of these two residues. From the sum of our data,we conclude that theblood group M and N specific determinants have three identical carbohydrate moieties per peptide. Therefore, the antigenic specificity has to be determined by differences in the amino acid sequence at residues 1 and/or 5. Since deamination of the Mspecific octapeptide results in loss of activity, we conclude that for this antigen the NHz-terminal serine (which includes a positive charge) provides the immunodominant and specific structure which is recognized by commercial rabbit antibodies. Our data thus do not support the views and interpretations put forward by Springer and his co-workers (10, 11), who postulated a biosynthetic precursor relationship between Tn, T, N, and M antigens. The MN antigens, as defined here, are not solely dependent on carbohydrate structures which possibly is true for T and Tn (1 1).
