Immunochemistry, 1975, Vol. 12, pp. 653 656. Pergamon Press. Printed in Great Britain
H U M A N BLOOD SUBUNIT COMPOSITION OF i G R O U P M N ANTIGENS* JOEL KIRSCHBAUM and GEORG F. SPRINGER Squibb Institute for Medical Research, New Brunswick, New Jersey, 08903, and Department of Immunochemistry Research, Evanston Hospital, and Department of Microbiology, Northwestern University, Evanston, Illinois, 60201, U.S.A. (First received 15 August 1974; in revised form 16 October 1974)
Abstract--The mol. wt of highly purified blood group MN antigens from human erythrocytes depends on the method of isolation and the nature of the solvent. The mol. wt of the active repeating subunits isolated by gentle and non-dissociating procedures (pH 6-0-7.2 and 23-25°C) was found to be 50,000 and 53,000 when dissolved in 9 M formamide and 5% ethanol-water, respectively.Isolation at pH > 7.8 or >50°C yielded preparations with a mol. wt of 30,000; these latter, smaller, only weakly active molecules may be degradation or disruption products of strong bound dimers. Comparison of 'native' with artificially acetylated antigen indicate that both hydrogen ,and hydrophobic bonding contribute to stabilize the aggregate.
INTRODUCTION Reports by Bezkorovainy et al. (1966) and Springer et al. (1966) detailed some physical, chemical and biological properties on the MN substances, the major antigens of the second human blood group system (Landsteiner and Levine, 1928). Highly active, homogeneous preparations were described (Bezkorovainy et al., 1966). These substances are glycoproteins with no significant amount of chloroform-methanol extractable lipid (Springer et al., 1966; Fukuda and Osawa, 1973; Blumenfeld et al., 1970). The antigens are also myxovirus receptors (Springer and Ansell, 1958; Kathan et al., 1961). Molecular weights were detero diffumined from the sedimentation constants S2o,w, sion constants O20. o w and the calculated partial specific volume ~ of 0'68, by the Svedberg equation (Schachman, 1959). Molecular weights from 1.5 x 105 to 12 x 106 have been found for these substances (Bezkorovainy et al., 1966; Springer, 1967), which are apparently composed of identical subunits (Kathan et al., 1961; Springer et al., 1966; Springer, 1967) and were believed to have a mol. wt. of ca. 30,000. The glycoproteins possessed rather high frictional ratios which were indicative of substantial molecular asymmetry (Bezkorovainy et al., 1966). No physical differences were found between M and N antigens. However, substances of 30,000 mol. wt were present in physiological solvents only when rather harsh isolation procedures, such as with hot aqueous phenol under alkaline conditions (Kathan et al., 1961; Morawiecki, 1964) had been employed. The blood group and anti-viral activities of such harshly isolated preparations were faint (Kathan et al., 1961; Springer et al., 1966; Springer, 1967; Springer et al., 1969). The MN glycoproteins of much larger mol. wt extracted under mild conditions had higher conformational order, as determined by optical methods (Jirgensons and Springer, 1968) and were over a thousand times more active on a
molar basis than these subunits (Springer, 1967; Springer et al., 1969). A model has been proposed which stresses the existence in asymmetric arrangement of hydrophilic and hydrophobic groupings on the M and N antigens (Morawiecki, 1964). In this model the MN antigens were envisioned to occur in nature only as substances with a mol. wt of 30,000; each of these molecules was thought to be held only via hydrophobic bonds in a lipidic layer of the erythrocyte membrane. The above cited observations on the units of 30,000 mol. wt and indication of the importance of bonds in addition to apolar ones for the molecular arrangement of the substances (Springer e t al., 1966; Jirgensons and Springer, 1968), prompted the present investigation. We confirmed that the blood group MN antigens are composed of subunits but arrived at a molecular size for the basic subunit which was different from that proposed earlier, provided neutral pH and low temperatures prevailed. A molecular size similar to that suggested recently by Marchesi et al. (1972) and Fukuda and Osawa (1973) was found. The present experiments also shed new light on the types of bonding that stabilize the aggregated units. MATERIALS AND METHODS
Isolation and acetylation of antigens Reagent grade or equivalent chemicals were used throughout. The blood group MM and NN antigens were obtained from donors of blood group O homozygous for either M or N and purified as described previously; the various preparations used in the present study corresponded to those designated earlier as 'purified' (Springer et al., 1966). Usually the sediments were dissolved in water containing 0.1% sodium acetate and fractionated with ethanol. The most highly blood group active fractions which were precipitated by ethanol concentrations of 50-75 per cent were used. The sialic acid content of these preparations ranged from 10.0 to 21.3 per cent as determined by thiobar* This work was supported in part by grant AI-05681 bituric acid test (cf. Warren, 1959) and they contained 7.9from the United States Public Health Service and G. F. S. 11.4% hexosamine (as galactosamine) in the assay of Gatt by the Dr. William R. Parkes Fund for Cancer Research. and Berman (1966). 653
654
JOEL KIRSCHBAUM and GEORG F. SPRINGER
'Native' as well as acetylated antigens were employed. Acetylation was with acetic anhydride and based on the mild procedure of Prey and Aszalos (1960) in which no degradation and rearrangements have been observed. The preparations were thoroughly electrodialyzed and dried to constant weight at 23-25°C (cf. Springer et al., 1966). Approximately 100 mg of antigen were dissolved overnight at 4°C in 20 ml distilled water in a beaker and stirred magnetically. Excess acetic anhydride in ethyl ether was added dropwise from a buret; the pH was kept between 7.0 and 8.0 by the addition of 0-01 N NaOH from a second buret as monitored with a Coleman Metrion III pH meter. Preliminary experiments showed that an acetic anhydridecarbohydrate molar ratio of 50 : 1 yielded the most extensive acetylation. The carbohydrate figure was based on moles of component monosaccharide/mole of antigen employed. Appropriate blanks and controls that measured acetic anhydride consumption were included. At the end of the reaction, the antigen was thoroughly dialyzed, electrodialyzed, freeze-dried and dried to a constant weight at 2325°C. Acetylation was repeated twice under the above conditions. Thereafter no significant increase in acetyl content of the antigen was observed by the procedure of LudowiegDorfman (1960) in which methyl acetate was used as standard. The final product was checked for free acetyl by gas chromatography (Nagal and Watanabc, 1969); < 0.5 #mole of free acetyl/100mg of antigen was found. The concentration of sialic acid, neutral sugars expressed as galactose as well as hexosamine were also measured as described earlier (Springer et al., 1966). Blood group and myxovirus activities were determined as described by Springer et al. (1966). O,MM antigen UC58 'native' as well as acetylated was employed extensively in the present study. Its carbohydrate composition and its biological activities are listed in Table I. The acetyl content of the untreated antigen (weight basis) amounted to 110.6 per cent of the theory assuming mono-N-acetylation of the hexosamines and sialic acid. The values are higher than theory since the MN antigens contain small quantities of N,O-di-acetyl neuraminic acid (Springer et al., 1966). Blood group and anti-viral activities of the 'native' UC58 antigen were approximately one-tenth of those of the most highly active preparations but exceeded ca. 50-fold those of 30,000 tool. wt (Springer, 1967; Springer et al., 1969). After in vitro acetylation, the total acetyl content was 12.24 per cent. This is 71.2 per cent of theory assuming an average of three accessible-OH groups per monosaccharide unit of MM antigen. There was no loss of carbohydrate considering the increase in acetyl content of the molecule. However, the blood group
M activity was almost completely destroyed and the influenza virus inhibitory power reduced between 75 and 90 per cent (Table 1). Physical procedures
Sedimentation coefficients were determined in a Spinco Model E analytical ultracentrifuge at 5 or 20°C, with the use of a calibrated temperature-control unit. Sedimentation coefficients were corrected to standard conditions, namely the viscosity and density of water at 20°C (cf. Schachman, 1959). In the disrupting solvents, viscosity and density corrections taken from the data of Kawahara and Tanford (1966) were applied. Mol. wts were determined by the Archibald (1947) approach-to-sedimentation equilibrium method because the rapidity of the procedure minimized the possibility for reaggregation of the subunits (Millar et al., 1960). Unless stated otherwise, antigens were analyzed at a concentration of 5 m g / m l in a 12-mm, double sector interference cell. The concentration gradients at the meniscus and cell bottom were measured from data recorded on Kodak Metallographic plates magnified ten-fold in a Nikon magnifier. Total concentration was measured with the aid of a capillary centerpiece. The integral of the concentration gradient was evaluated by the procedure of Engelberg (1963). The partial specific volume of 0'68 was decreased by 0.01 cm3/g for calculation of tool. wt when buffered 8 M urea, 9 M formamide or 5 M guanidinium chloride was used as solvent (Tanford et al., 1967). Before measurement all solutions were equilibrated at 23°C for 1 hr.
RESULTS AND DISCUSSION Antigens cxisted as aggregates in aqueous solvents as shown by their sedimentation coefficient (s) which decreased with decreasing antigen concentration (c). The relationship between s a n d c was linear with d s / d c = - 1"5 × 10-14. Sedimentation coefficients were determined in 0 . 2 M sodium chloride - 0 . 2 M phosphate buffer p H 6.85 o n several different preparations, which were usually polydisperse. A wide range of S2o.w was found for some preparations. Figure I shows the Schlieren pattern of a large molecular weight preparation which o n dissolution in 5 . 0 M aqueous NaC1 h a d a n S2o,w of 60. Table 2 lists the extremes observed. Occasionally, the sedim e n t a t i o n coefficient approached 2000. N o t more than two preparations were examined in each of the
Table 1. Carbohydrate composition and biological properties of blood group 0,MM antigens UC58 before and after acetylation in vitro
Inhibitory A~tlvltiera Antigen
Sialic Acid
Hexosamlnes
Neutral Sugars
Z
~
~
Acetyl Myxovirus
Z
Not Acetylated
12.20
9.93
8.05
4.05
Acetylated
11.15
9.09
7.92
12.24
Anti-M
0.04 ±2.5
A/PR 8
s/~
0.002
0.02
0.02
0.06
Concentration (mg/ml) completely inhibiting the agglutination of human 0,~4 erythrocytes by 4 doses of 2 different htmmn anti-M sera and the A/PR 8 and B/MD indicator influenza virus stralns, respectively (cf. Springer etal., 1966).
Subunits of Blood Group MN Antigens
Fig. 1. Ultracentrifugal analysis of human blood group MM glycoprotein at a concentration of 5 mg/ml in 5.0 M aqueous NaCI. Picture was taken 20 min after reaching speed at 6995 rev/min, 20°C and bar angle of 80°. Sedimentation to the right.
other solvents in which the apparent mol. wt depended on the nature of the solvent. One molecular species predominated by >75 per cent except in guanidinium chloride where two species were found in about equal proportions (Table 2). In addition, the sedimentation coefficient of the antigens was dependent on speed of rotation. Thus for one preparation at 0'66Ve concentration s in 0"2 M NaCI-0.02M phosphate buffer pH6"85 was 15 at 12,590 rev/min and 20 at 42,040 rev/min.
655
Electrostatic forces may have only a minor role in stabilization of the aggregates, since solvents at acid pH or of high ionic strength did not yield antigen solutions with low s values. The sedimentation data at alkaline pH are suspect, since degradation of the glycoproteins, occurred, as indicated by earlier studies (Springer et al., 1966; Jirgensons and Springer, 1968). Table 2 shows that in 8 M urea and 1~o sodium dodecyl sulfate the apparent mol. wt of the antigen was approximately 300,000 daltons, uncorrected for SDS binding and in 5 M guanidinium chloride the apparent mol. wt was 86,000. A second species was present with a mol. wt of 190,000, which most likely represents a dimer of the major component. Disaggregation in guanidinium chloride was not complete, since in 9 M formamide the antigen had a mol. wt of 53,000, and in 5~o ethanol-water it was 50,000. These latter molecular weights are in accord with recent findings by others (Marchesi et al., 1973; Fukuda and Osawa, 1973). The species reported earlier to result after formamide treatment (Morawiecki, 1964) may be a dimer. Table 3 compares the s20,w values in water and in aqueous 5~ alcohol solutions of the artificially acetylated antigen with the 'native' product to assess the effect of hydroxyl groups on the aggregation of subunits. Stabilization of the aggregate by means of hydrogen bonding is indicated by the higher sedimentation coefficients of 'native' compared to artificially acetylated antigen in water and in aqueous methanol (Table 3). Lower s values for the artificially acetylated antigen in these two solvents indicate that the blocking of hydroxyl groups results in less aggregation. Evidence for the presence of hydrophobic bonding may be seen in the lesser disruption of artificially acetylated antigen, as compared to the 'native' preparation, in aqueous ethanol and butanol. The 5~o ethanol-water solvent system was the most effective of the
Table 2. Sedimentation coefficients and apparent molecular weights of M and N antigens in various solvents
Solvent
s
-20,w
Molecular Weight
0.2 M NaCl - 0.02 M phosphate buffer, pH 6.85 (P)~ 5 M NaC1
18 - 1980 60
0.05 M Glyelne - HC1 buffer, pH 2.2 0.2 M Glycine NaOH buffer, pH 10.4
25 6,1
8 M Urea in P
290,000 ± 5,000
1% Sodium dodecyl sulfate in P
280,000 ± 5,000
5 M Guanidinlum chlorlde in P
86,000; 190,000
9 M Formamide in P
53,000 ± 5,000
5% Ethanol - 95% water
50,000 ± 5,000
P = 0.2 M NaCl - 0.02 M phosphate buffer, pH 6.85.
656
JOEL KIRSCHBAUM and GEORG F. SPRINGER
Table 3. s20,w of native and acetylated 0,MM antigen UC58 in water and water-alcohol mixtures
L20,w Solvent
Not artlfieially acetylated
Water
17, 64
5% Methanol - 95% w a t e r
73, 148
Aeetylated
Ii 8.7
5% Ethanol - 95% water
4.4
7.1
5% n-Propanol - 95% w a t e r
6.2
6.2
5% n-Butanol - 95% w a t e r
6.9
8.5
five alcohol-water solvents tested in the disaggregation of 'native' antigen (Table 3). If hydrogen bonding were more important than hydrophobic interaction in stabilizing the aggregate, then aqueous methanol should be the most effective disaggregating solvent. If hydrophobic bonding were more important, then aqueous butanol should be most effective (Kirschbaum et al., 1970). The disaggregation efficacy of propanol, an alcohol of intermediate hydrocarbon chain length, indicated that both kinds of bonding contribute. The subunit with a mol. wt of 30,000 was obtained by us only when we isolated the M N substances at temperatures above 50°C or above pH 8.0. It is likely to be a fragmented product of the subunit with a mol. wt of 53,000. Our present experiments, like earlierones by others (Marchesi et al., 1972) do not clarify whether the M N antigens occur in the red cell membrane as individual subunits or as aggregates. However, we have never been able to isolate under mild conditions with non-dissociating solvents the subunits which are of low conformational order when compared to the extensively aggregated glycoproteins as determined with circular dichroic and optical rotatory dispersion measurements (Jirgensons and Springer, 1968). Also, the subunits mol. wt 50,000 possessed only about 5 per cent of the blood group and myxovirus inhibitory activity of the not artificially acetylated M M antigen listed in Table 1. We, therefore, favor the assumption that, in addition t o subunits, chains or webs of subunits anchored in the lipidic part of the cell membrane, are spread over the red cell surface. Acknowledgements--We thank Mr. A. Pudzianowski, B. S.
and Mrs. H. Tegtmeyer, M. T., for expert technical help, and Dr. J. Dunham, Dr. D. Frost, Mr. H. Kadin and Dr. R. Millonig for their helpful comments and suggestions. REFERENCES
Archibald W. J. (1947) J. phys. Colloid Chem. 51, 1204. Bezkorovainy A., Springer G. F. and Hotta K. (1966) Biochim. biophys. Acta 115, 501. Blumenfeld O. O., Gallop P. M., Howe C. and Lee L. TI (1970) Biochim. biophys. Acta 211, 109. Engelberg J. (1963) Analyt. Biochem. 6, 530. Fukuda M. and Osawa T. (1973) J. biol. Chem. 248. 5100.
Gatt R. and Berman E. R. (1966) Analyt. Biochem. 15, 167. Jirgensons B. and Springer G. F. (1968) Science 162, 365. Kathan R. H., Winzler R. J. and Johnson C. A. (1961) J. exp. Med. 113, 37. Kawahara K. and Tanford C. (1966) J. biol. Chem. 241, 3228. Kirschbaum J., Slusarchyk W. A. and Weisenborn F. C. (1970) J. pharm. Sci. 59, 749. Landsteiner K. and Levine P. (1928) J. exp. Med. 47, 757. Lud0wieg J. and Dorfman A. (1960) Biochim. biophys. Acta 38, 212. Marchesi V. T., Tillack T. W., Jackson R. L., Segrest J. P. and Scott R. E. (1972) Proc. natn. Acad. Sci. U.S.A. 69, 1445. Millar D. B. S., Willick G. E., Steiner R. F. and Frattali V. (1969) J. biol. Chem. 244, 281. Morawiecki A. (1964) Biochim. biophys. Acta 83, 339. Nagai Y. and Watanabe T. (1969) Fukushima J. reed. Sci. 16, 115. Prey V. and Aszalos A. (1960) Mh. Chem. 91, 729. Schachman H. K. (1959) Ultracentrifugation in Biochemistry, p. 81. Academic Press, New York. Springer G. F. (1967) Biochem. Biophys. Res. Commun. 28, 510. Springer G. F. and Ansell N. J. (1958) Proc. natn. Acad. Sci. U.S.A. 44, 182. Springer G. F., Nagai Y. and Tegtmeyer H. (1966) Biochemistry 5, 3254. Springer G. F., Schwick H. G. and Fletcher M. A. (1969) Proc. natn. Acad. Sci. U.S.A. 64, 634. Tanford C., Kawahara K. and Lapanje S. (1967) J. Am. chem. Soc. 89, 729. Warren L. (1959) J. biol. Chem. 234, 1971.
NOTE A D D E D IN PROOF
After this paper was in press, Grefrath S. P. and Reynolds J. A. (1974) Proc. natn. Acad. Sci. U.S.A. 71, 3913 reported the mol. wt of the glycoprotein described above to be 29,000 after dissolution in SDS, heating to 80°C and removal of all SDS. We treated M and N glycoproteins as these authors. A preparation treated with SDS and heated as well as a control preparation treated alike but without SDS and heating had both s 7.6 after 6 days dialysis when dissolved in the NaCI-PO 4 buffer of pH 6.85 determined by the sedimentation equilibrium procedure described above. However, when compared to the control the activity of the SDS-treated, heated glycoprotein was destroyed by 90-99~o as measured with 2 human anti-M sera and A/PR8 influenza virus and to 50-75~o when measured with rabbit anti-M and Vicia reagent.
H U M A N BLOOD SUBUNIT COMPOSITION OF i G R O U P M N ANTIGENS* JOEL KIRSCHBAUM and GEORG F. SPRINGER Squibb Institute for Medical Research, New Brunswick, New Jersey, 08903, and Department of Immunochemistry Research, Evanston Hospital, and Department of Microbiology, Northwestern University, Evanston, Illinois, 60201, U.S.A. (First received 15 August 1974; in revised form 16 October 1974)
Abstract--The mol. wt of highly purified blood group MN antigens from human erythrocytes depends on the method of isolation and the nature of the solvent. The mol. wt of the active repeating subunits isolated by gentle and non-dissociating procedures (pH 6-0-7.2 and 23-25°C) was found to be 50,000 and 53,000 when dissolved in 9 M formamide and 5% ethanol-water, respectively.Isolation at pH > 7.8 or >50°C yielded preparations with a mol. wt of 30,000; these latter, smaller, only weakly active molecules may be degradation or disruption products of strong bound dimers. Comparison of 'native' with artificially acetylated antigen indicate that both hydrogen ,and hydrophobic bonding contribute to stabilize the aggregate.
INTRODUCTION Reports by Bezkorovainy et al. (1966) and Springer et al. (1966) detailed some physical, chemical and biological properties on the MN substances, the major antigens of the second human blood group system (Landsteiner and Levine, 1928). Highly active, homogeneous preparations were described (Bezkorovainy et al., 1966). These substances are glycoproteins with no significant amount of chloroform-methanol extractable lipid (Springer et al., 1966; Fukuda and Osawa, 1973; Blumenfeld et al., 1970). The antigens are also myxovirus receptors (Springer and Ansell, 1958; Kathan et al., 1961). Molecular weights were detero diffumined from the sedimentation constants S2o,w, sion constants O20. o w and the calculated partial specific volume ~ of 0'68, by the Svedberg equation (Schachman, 1959). Molecular weights from 1.5 x 105 to 12 x 106 have been found for these substances (Bezkorovainy et al., 1966; Springer, 1967), which are apparently composed of identical subunits (Kathan et al., 1961; Springer et al., 1966; Springer, 1967) and were believed to have a mol. wt. of ca. 30,000. The glycoproteins possessed rather high frictional ratios which were indicative of substantial molecular asymmetry (Bezkorovainy et al., 1966). No physical differences were found between M and N antigens. However, substances of 30,000 mol. wt were present in physiological solvents only when rather harsh isolation procedures, such as with hot aqueous phenol under alkaline conditions (Kathan et al., 1961; Morawiecki, 1964) had been employed. The blood group and anti-viral activities of such harshly isolated preparations were faint (Kathan et al., 1961; Springer et al., 1966; Springer, 1967; Springer et al., 1969). The MN glycoproteins of much larger mol. wt extracted under mild conditions had higher conformational order, as determined by optical methods (Jirgensons and Springer, 1968) and were over a thousand times more active on a
molar basis than these subunits (Springer, 1967; Springer et al., 1969). A model has been proposed which stresses the existence in asymmetric arrangement of hydrophilic and hydrophobic groupings on the M and N antigens (Morawiecki, 1964). In this model the MN antigens were envisioned to occur in nature only as substances with a mol. wt of 30,000; each of these molecules was thought to be held only via hydrophobic bonds in a lipidic layer of the erythrocyte membrane. The above cited observations on the units of 30,000 mol. wt and indication of the importance of bonds in addition to apolar ones for the molecular arrangement of the substances (Springer e t al., 1966; Jirgensons and Springer, 1968), prompted the present investigation. We confirmed that the blood group MN antigens are composed of subunits but arrived at a molecular size for the basic subunit which was different from that proposed earlier, provided neutral pH and low temperatures prevailed. A molecular size similar to that suggested recently by Marchesi et al. (1972) and Fukuda and Osawa (1973) was found. The present experiments also shed new light on the types of bonding that stabilize the aggregated units. MATERIALS AND METHODS
Isolation and acetylation of antigens Reagent grade or equivalent chemicals were used throughout. The blood group MM and NN antigens were obtained from donors of blood group O homozygous for either M or N and purified as described previously; the various preparations used in the present study corresponded to those designated earlier as 'purified' (Springer et al., 1966). Usually the sediments were dissolved in water containing 0.1% sodium acetate and fractionated with ethanol. The most highly blood group active fractions which were precipitated by ethanol concentrations of 50-75 per cent were used. The sialic acid content of these preparations ranged from 10.0 to 21.3 per cent as determined by thiobar* This work was supported in part by grant AI-05681 bituric acid test (cf. Warren, 1959) and they contained 7.9from the United States Public Health Service and G. F. S. 11.4% hexosamine (as galactosamine) in the assay of Gatt by the Dr. William R. Parkes Fund for Cancer Research. and Berman (1966). 653
654
JOEL KIRSCHBAUM and GEORG F. SPRINGER
'Native' as well as acetylated antigens were employed. Acetylation was with acetic anhydride and based on the mild procedure of Prey and Aszalos (1960) in which no degradation and rearrangements have been observed. The preparations were thoroughly electrodialyzed and dried to constant weight at 23-25°C (cf. Springer et al., 1966). Approximately 100 mg of antigen were dissolved overnight at 4°C in 20 ml distilled water in a beaker and stirred magnetically. Excess acetic anhydride in ethyl ether was added dropwise from a buret; the pH was kept between 7.0 and 8.0 by the addition of 0-01 N NaOH from a second buret as monitored with a Coleman Metrion III pH meter. Preliminary experiments showed that an acetic anhydridecarbohydrate molar ratio of 50 : 1 yielded the most extensive acetylation. The carbohydrate figure was based on moles of component monosaccharide/mole of antigen employed. Appropriate blanks and controls that measured acetic anhydride consumption were included. At the end of the reaction, the antigen was thoroughly dialyzed, electrodialyzed, freeze-dried and dried to a constant weight at 2325°C. Acetylation was repeated twice under the above conditions. Thereafter no significant increase in acetyl content of the antigen was observed by the procedure of LudowiegDorfman (1960) in which methyl acetate was used as standard. The final product was checked for free acetyl by gas chromatography (Nagal and Watanabc, 1969); < 0.5 #mole of free acetyl/100mg of antigen was found. The concentration of sialic acid, neutral sugars expressed as galactose as well as hexosamine were also measured as described earlier (Springer et al., 1966). Blood group and myxovirus activities were determined as described by Springer et al. (1966). O,MM antigen UC58 'native' as well as acetylated was employed extensively in the present study. Its carbohydrate composition and its biological activities are listed in Table I. The acetyl content of the untreated antigen (weight basis) amounted to 110.6 per cent of the theory assuming mono-N-acetylation of the hexosamines and sialic acid. The values are higher than theory since the MN antigens contain small quantities of N,O-di-acetyl neuraminic acid (Springer et al., 1966). Blood group and anti-viral activities of the 'native' UC58 antigen were approximately one-tenth of those of the most highly active preparations but exceeded ca. 50-fold those of 30,000 tool. wt (Springer, 1967; Springer et al., 1969). After in vitro acetylation, the total acetyl content was 12.24 per cent. This is 71.2 per cent of theory assuming an average of three accessible-OH groups per monosaccharide unit of MM antigen. There was no loss of carbohydrate considering the increase in acetyl content of the molecule. However, the blood group
M activity was almost completely destroyed and the influenza virus inhibitory power reduced between 75 and 90 per cent (Table 1). Physical procedures
Sedimentation coefficients were determined in a Spinco Model E analytical ultracentrifuge at 5 or 20°C, with the use of a calibrated temperature-control unit. Sedimentation coefficients were corrected to standard conditions, namely the viscosity and density of water at 20°C (cf. Schachman, 1959). In the disrupting solvents, viscosity and density corrections taken from the data of Kawahara and Tanford (1966) were applied. Mol. wts were determined by the Archibald (1947) approach-to-sedimentation equilibrium method because the rapidity of the procedure minimized the possibility for reaggregation of the subunits (Millar et al., 1960). Unless stated otherwise, antigens were analyzed at a concentration of 5 m g / m l in a 12-mm, double sector interference cell. The concentration gradients at the meniscus and cell bottom were measured from data recorded on Kodak Metallographic plates magnified ten-fold in a Nikon magnifier. Total concentration was measured with the aid of a capillary centerpiece. The integral of the concentration gradient was evaluated by the procedure of Engelberg (1963). The partial specific volume of 0'68 was decreased by 0.01 cm3/g for calculation of tool. wt when buffered 8 M urea, 9 M formamide or 5 M guanidinium chloride was used as solvent (Tanford et al., 1967). Before measurement all solutions were equilibrated at 23°C for 1 hr.
RESULTS AND DISCUSSION Antigens cxisted as aggregates in aqueous solvents as shown by their sedimentation coefficient (s) which decreased with decreasing antigen concentration (c). The relationship between s a n d c was linear with d s / d c = - 1"5 × 10-14. Sedimentation coefficients were determined in 0 . 2 M sodium chloride - 0 . 2 M phosphate buffer p H 6.85 o n several different preparations, which were usually polydisperse. A wide range of S2o.w was found for some preparations. Figure I shows the Schlieren pattern of a large molecular weight preparation which o n dissolution in 5 . 0 M aqueous NaC1 h a d a n S2o,w of 60. Table 2 lists the extremes observed. Occasionally, the sedim e n t a t i o n coefficient approached 2000. N o t more than two preparations were examined in each of the
Table 1. Carbohydrate composition and biological properties of blood group 0,MM antigens UC58 before and after acetylation in vitro
Inhibitory A~tlvltiera Antigen
Sialic Acid
Hexosamlnes
Neutral Sugars
Z
~
~
Acetyl Myxovirus
Z
Not Acetylated
12.20
9.93
8.05
4.05
Acetylated
11.15
9.09
7.92
12.24
Anti-M
0.04 ±2.5
A/PR 8
s/~
0.002
0.02
0.02
0.06
Concentration (mg/ml) completely inhibiting the agglutination of human 0,~4 erythrocytes by 4 doses of 2 different htmmn anti-M sera and the A/PR 8 and B/MD indicator influenza virus stralns, respectively (cf. Springer etal., 1966).
Subunits of Blood Group MN Antigens
Fig. 1. Ultracentrifugal analysis of human blood group MM glycoprotein at a concentration of 5 mg/ml in 5.0 M aqueous NaCI. Picture was taken 20 min after reaching speed at 6995 rev/min, 20°C and bar angle of 80°. Sedimentation to the right.
other solvents in which the apparent mol. wt depended on the nature of the solvent. One molecular species predominated by >75 per cent except in guanidinium chloride where two species were found in about equal proportions (Table 2). In addition, the sedimentation coefficient of the antigens was dependent on speed of rotation. Thus for one preparation at 0'66Ve concentration s in 0"2 M NaCI-0.02M phosphate buffer pH6"85 was 15 at 12,590 rev/min and 20 at 42,040 rev/min.
655
Electrostatic forces may have only a minor role in stabilization of the aggregates, since solvents at acid pH or of high ionic strength did not yield antigen solutions with low s values. The sedimentation data at alkaline pH are suspect, since degradation of the glycoproteins, occurred, as indicated by earlier studies (Springer et al., 1966; Jirgensons and Springer, 1968). Table 2 shows that in 8 M urea and 1~o sodium dodecyl sulfate the apparent mol. wt of the antigen was approximately 300,000 daltons, uncorrected for SDS binding and in 5 M guanidinium chloride the apparent mol. wt was 86,000. A second species was present with a mol. wt of 190,000, which most likely represents a dimer of the major component. Disaggregation in guanidinium chloride was not complete, since in 9 M formamide the antigen had a mol. wt of 53,000, and in 5~o ethanol-water it was 50,000. These latter molecular weights are in accord with recent findings by others (Marchesi et al., 1973; Fukuda and Osawa, 1973). The species reported earlier to result after formamide treatment (Morawiecki, 1964) may be a dimer. Table 3 compares the s20,w values in water and in aqueous 5~ alcohol solutions of the artificially acetylated antigen with the 'native' product to assess the effect of hydroxyl groups on the aggregation of subunits. Stabilization of the aggregate by means of hydrogen bonding is indicated by the higher sedimentation coefficients of 'native' compared to artificially acetylated antigen in water and in aqueous methanol (Table 3). Lower s values for the artificially acetylated antigen in these two solvents indicate that the blocking of hydroxyl groups results in less aggregation. Evidence for the presence of hydrophobic bonding may be seen in the lesser disruption of artificially acetylated antigen, as compared to the 'native' preparation, in aqueous ethanol and butanol. The 5~o ethanol-water solvent system was the most effective of the
Table 2. Sedimentation coefficients and apparent molecular weights of M and N antigens in various solvents
Solvent
s
-20,w
Molecular Weight
0.2 M NaCl - 0.02 M phosphate buffer, pH 6.85 (P)~ 5 M NaC1
18 - 1980 60
0.05 M Glyelne - HC1 buffer, pH 2.2 0.2 M Glycine NaOH buffer, pH 10.4
25 6,1
8 M Urea in P
290,000 ± 5,000
1% Sodium dodecyl sulfate in P
280,000 ± 5,000
5 M Guanidinlum chlorlde in P
86,000; 190,000
9 M Formamide in P
53,000 ± 5,000
5% Ethanol - 95% water
50,000 ± 5,000
P = 0.2 M NaCl - 0.02 M phosphate buffer, pH 6.85.
656
JOEL KIRSCHBAUM and GEORG F. SPRINGER
Table 3. s20,w of native and acetylated 0,MM antigen UC58 in water and water-alcohol mixtures
L20,w Solvent
Not artlfieially acetylated
Water
17, 64
5% Methanol - 95% w a t e r
73, 148
Aeetylated
Ii 8.7
5% Ethanol - 95% water
4.4
7.1
5% n-Propanol - 95% w a t e r
6.2
6.2
5% n-Butanol - 95% w a t e r
6.9
8.5
five alcohol-water solvents tested in the disaggregation of 'native' antigen (Table 3). If hydrogen bonding were more important than hydrophobic interaction in stabilizing the aggregate, then aqueous methanol should be the most effective disaggregating solvent. If hydrophobic bonding were more important, then aqueous butanol should be most effective (Kirschbaum et al., 1970). The disaggregation efficacy of propanol, an alcohol of intermediate hydrocarbon chain length, indicated that both kinds of bonding contribute. The subunit with a mol. wt of 30,000 was obtained by us only when we isolated the M N substances at temperatures above 50°C or above pH 8.0. It is likely to be a fragmented product of the subunit with a mol. wt of 53,000. Our present experiments, like earlierones by others (Marchesi et al., 1972) do not clarify whether the M N antigens occur in the red cell membrane as individual subunits or as aggregates. However, we have never been able to isolate under mild conditions with non-dissociating solvents the subunits which are of low conformational order when compared to the extensively aggregated glycoproteins as determined with circular dichroic and optical rotatory dispersion measurements (Jirgensons and Springer, 1968). Also, the subunits mol. wt 50,000 possessed only about 5 per cent of the blood group and myxovirus inhibitory activity of the not artificially acetylated M M antigen listed in Table 1. We, therefore, favor the assumption that, in addition t o subunits, chains or webs of subunits anchored in the lipidic part of the cell membrane, are spread over the red cell surface. Acknowledgements--We thank Mr. A. Pudzianowski, B. S.
and Mrs. H. Tegtmeyer, M. T., for expert technical help, and Dr. J. Dunham, Dr. D. Frost, Mr. H. Kadin and Dr. R. Millonig for their helpful comments and suggestions. REFERENCES
Archibald W. J. (1947) J. phys. Colloid Chem. 51, 1204. Bezkorovainy A., Springer G. F. and Hotta K. (1966) Biochim. biophys. Acta 115, 501. Blumenfeld O. O., Gallop P. M., Howe C. and Lee L. TI (1970) Biochim. biophys. Acta 211, 109. Engelberg J. (1963) Analyt. Biochem. 6, 530. Fukuda M. and Osawa T. (1973) J. biol. Chem. 248. 5100.
Gatt R. and Berman E. R. (1966) Analyt. Biochem. 15, 167. Jirgensons B. and Springer G. F. (1968) Science 162, 365. Kathan R. H., Winzler R. J. and Johnson C. A. (1961) J. exp. Med. 113, 37. Kawahara K. and Tanford C. (1966) J. biol. Chem. 241, 3228. Kirschbaum J., Slusarchyk W. A. and Weisenborn F. C. (1970) J. pharm. Sci. 59, 749. Landsteiner K. and Levine P. (1928) J. exp. Med. 47, 757. Lud0wieg J. and Dorfman A. (1960) Biochim. biophys. Acta 38, 212. Marchesi V. T., Tillack T. W., Jackson R. L., Segrest J. P. and Scott R. E. (1972) Proc. natn. Acad. Sci. U.S.A. 69, 1445. Millar D. B. S., Willick G. E., Steiner R. F. and Frattali V. (1969) J. biol. Chem. 244, 281. Morawiecki A. (1964) Biochim. biophys. Acta 83, 339. Nagai Y. and Watanabe T. (1969) Fukushima J. reed. Sci. 16, 115. Prey V. and Aszalos A. (1960) Mh. Chem. 91, 729. Schachman H. K. (1959) Ultracentrifugation in Biochemistry, p. 81. Academic Press, New York. Springer G. F. (1967) Biochem. Biophys. Res. Commun. 28, 510. Springer G. F. and Ansell N. J. (1958) Proc. natn. Acad. Sci. U.S.A. 44, 182. Springer G. F., Nagai Y. and Tegtmeyer H. (1966) Biochemistry 5, 3254. Springer G. F., Schwick H. G. and Fletcher M. A. (1969) Proc. natn. Acad. Sci. U.S.A. 64, 634. Tanford C., Kawahara K. and Lapanje S. (1967) J. Am. chem. Soc. 89, 729. Warren L. (1959) J. biol. Chem. 234, 1971.
NOTE A D D E D IN PROOF
After this paper was in press, Grefrath S. P. and Reynolds J. A. (1974) Proc. natn. Acad. Sci. U.S.A. 71, 3913 reported the mol. wt of the glycoprotein described above to be 29,000 after dissolution in SDS, heating to 80°C and removal of all SDS. We treated M and N glycoproteins as these authors. A preparation treated with SDS and heated as well as a control preparation treated alike but without SDS and heating had both s 7.6 after 6 days dialysis when dissolved in the NaCI-PO 4 buffer of pH 6.85 determined by the sedimentation equilibrium procedure described above. However, when compared to the control the activity of the SDS-treated, heated glycoprotein was destroyed by 90-99~o as measured with 2 human anti-M sera and A/PR8 influenza virus and to 50-75~o when measured with rabbit anti-M and Vicia reagent.
